BEGIN:VCALENDAR VERSION:2.0 PRODID:researchseminars.org CALSCALE:GREGORIAN X-WR-CALNAME:researchseminars.org BEGIN:VEVENT SUMMARY:Christophe Biscio (Aalborg University) DTSTART:20220929T131500Z DTEND:20220929T140000Z DTSTAMP:20260422T122728Z UID:gbgstats/1 DESCRIPTION:Title: Asymptotic topological data analysis for point processes\nby Christop he Biscio (Aalborg University) as part of Gothenburg statistics seminar\n\ nLecture held in MVL14.\n\nAbstract\nTopological Data Analysis has in the past year attracted more attention in various fields such as in material s ciences to study the properties of porous material or in statistics to stu dy the asymptotic properties of random objects. However\, topological data analysis still appears hard to grasp for many statisticians. \n\nThis tal k intends to be an introduction to topological data analysis and therefore does not require any background in the field. We will present an overview of the different approaches in topological data analysis and will focus o n the persistent homology approach. \nWe will present the framework of thi s approach and its main mathematical objects. \nFinally\, we come back to the land of Probability and will present a central limit theorem for the s o-called Betti numbers obtained from stationary point processes\, non-nece ssarily Poisson.\n LOCATION:https://researchseminars.org/talk/gbgstats/1/ END:VEVENT BEGIN:VEVENT SUMMARY:Anders Ståhlberg & Serik Sagitov (Chalmers & University of Gothen burg) DTSTART:20221006T131500Z DTEND:20221006T140000Z DTSTAMP:20260422T122728Z UID:gbgstats/2 DESCRIPTION:Title: Counting molecular identifiers in sequencing using a multitype branching process with immigration\nby Anders Ståhlberg & Serik Sagitov (Chalme rs & University of Gothenburg) as part of Gothenburg statistics seminar\n\ nLecture held in MVL14.\n\nAbstract\nDetection of extremely rare variant a lleles\, such as tumour DNA\, within a complex mixture of DNA molecules is experimentally challenging due to sequencing errors. Barcoding of target DNA molecules in library construction for next-generation sequencing provi des a way to identify and bioinformatically remove polymerase induced erro rs. During the barcoding procedure involving $t$ consecutive PCR cycles\, the DNA molecules become barcoded by unique molecular identifiers (UMI). D ifferent library construction protocols utilise different values of $t$. T he effect of a larger $t$ and imperfect PCR amplifications is poorly descr ibed. \n\nThis paper proposes a branching process with growing immigration as a model describing the random outcome of $t$ cycles of PCR barcoding . Our model discriminates between five different amplification rates $r_1$ \, $r_2$\, $r_3$\, $r_4$\, $r$ for different types of molecules associated with the PCR barcoding procedure. We study this model by focussing on $C_ t$\, the number of clusters of molecules sharing the same \nUMI\, as well as $C_t(m)$\, the number of UMI clusters of size $m$. Our main finding i s a remarkable asymptotic pattern valid for moderately large $t$. It turns out that \n$E(C_t(m))/E(C_t)\\approx 2^{-m}$ for $m=1\,2\,\\ldots$\, rega rdless of the underlying parameters $(r_1\,r_2\,r_3\,r_4\,r)$. The knowled ge of the quantities $C_t$ and $C_t(m)$ as functions of the experimental p arameters $t$ and $(r_1\,r_2\,r_3\,r_4\,r)$ will help the users to draw mo re adequate conclusions from the outcomes of different sequencing protocol s.\n LOCATION:https://researchseminars.org/talk/gbgstats/2/ END:VEVENT BEGIN:VEVENT SUMMARY:Peter Guttorp (University of Washington/Norwegian computing center ) DTSTART:20221027T131500Z DTEND:20221027T140000Z DTSTAMP:20260422T122728Z UID:gbgstats/3 DESCRIPTION:Title: Comparing recent climate models to data\nby Peter Guttorp (University of Washington/Norwegian computing center) as part of Gothenburg statistic s seminar\n\nLecture held in MVL14.\n\nAbstract\nThe latest climate model intercomparison project (CMIP6) was the basis for the sixth assessment rep ort of the Intergovernmental Panel on Climate Change. The design of CMIP6 included climate runs with historical forcings\, meant to be comparable to observational data. We will focus on global annual mean temperature\, a c ommon (but not particularly sensitive) measure of climate change. Using fo ur observational products provided with uncertainty assessments\, we combi ne these into a single series. In doing so\, we estimate a smooth trend an d a residual spectral density function\, with attendant simultaneous confi dence bands. Using the same kind of decomposition of 318 climate model run s from 58 models in the historical CMIP6 experiment\, we see how well the model runs agree with the data. We also compare the warming between 1880-1 899 and 1995-2014. This is joint work with Peter Craigmile of the Ohio Sta te University.\n LOCATION:https://researchseminars.org/talk/gbgstats/3/ END:VEVENT BEGIN:VEVENT SUMMARY:Olle Häggström (Chalmers University of Technology & University o f Gothenburg) DTSTART:20221013T131500Z DTEND:20221013T140000Z DTSTAMP:20260422T122728Z UID:gbgstats/4 DESCRIPTION:Title: Anthropic reasoning and the hinge of history hypothesis\nby Olle Häg gström (Chalmers University of Technology & University of Gothenburg) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstrac t\nAmong researchers and scholars working on existential risk and the long -term future of humanity\, it has become increasingly common to speak of o ur present time as uniquely pivotal for the long-term future - a notion th at has become known under the term Hinge of History (HoH). Recently\, atte mpts have been made to formalize this concept and work out whether we real ly do live during the HoH. This involves not only a Bayesian analysis but also controversial ideas in anthropic reasoning. I will review and critiqu e this work and arrive at a nuanced answer to the HoH question and its pra ctical ramifications.\n LOCATION:https://researchseminars.org/talk/gbgstats/4/ END:VEVENT BEGIN:VEVENT SUMMARY:Moritz Schauer (Chalmers University of Technology & University of Gothenburg) DTSTART:20221020T131500Z DTEND:20221020T140000Z DTSTAMP:20260422T122728Z UID:gbgstats/5 DESCRIPTION:Title: Automatic differentiation of programs with discrete randomness\nby Mo ritz Schauer (Chalmers University of Technology & University of Gothenburg ) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAb stract\nAutomatic differentiation (AD)\, a technique for constructing new programs which compute the derivative of an original program\, has become ubiquitous throughout scientific computing and deep learning due to the im proved performance afforded by gradient-based optimization. However\, AD s ystems have been restricted to the subset of programs that have a continuo us dependence on parameters. Programs that have discrete stochastic behavi ors governed by distribution parameters\, such as flipping a coin with pro bability p of being heads\, pose a challenge to these systems because the connection between the result (heads vs tails) and the parameters (p) is f undamentally discrete. In this paper we develop a new reparameterization-b ased methodology that allows for generating programs whose expectation is the derivative of the expectation of the original program. We showcase how this method gives an unbiased and low-variance estimator which is as auto mated as traditional AD mechanisms. We demonstrate unbiased forward-mode A D of discrete-time Markov chains\, agent-based models such as Conway's Gam e of Life\, and unbiased reverse-mode AD of a particle filter. Our code is available at https://github.com/gaurav-arya/StochasticAD.jl\n LOCATION:https://researchseminars.org/talk/gbgstats/5/ END:VEVENT BEGIN:VEVENT SUMMARY:Johan Jonasson (Chalmers University and University of Gothenburg) DTSTART:20230119T141600Z DTEND:20230119T150000Z DTSTAMP:20260422T122728Z UID:gbgstats/6 DESCRIPTION:Title: Noise sensitivity/stability for deep Boolean neural nets\nby Johan Jo nasson (Chalmers University and University of Gothenburg) as part of Gothe nburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nA well-kno wn and ubiquitous property of neural net classifiers is that they can be f ooled into misclassifying some objects by changing the input in tiny ways that are indistinguishable for the human eye. These changes can be adversa rial\, but sometimes they can be just random noise. This makes it interest ing to ask if this property is something that almost all neural nets have and\, when they do\, why that is. There are good heuristic explanations\, but to prove mathematically rigorous results seems very difficult in gener al. Here we prove some first results on various toy models. We treat our q uestions within the framework of the established field of noise sensitivit y/stability. What we prove can roughly be stated as:\n \n\n LOCATION:https://researchseminars.org/talk/gbgstats/6/ END:VEVENT BEGIN:VEVENT SUMMARY:David Widmann (Uppsala University) DTSTART:20221124T141500Z DTEND:20221124T150000Z DTSTAMP:20260422T122728Z UID:gbgstats/7 DESCRIPTION:Title: Calibration of probabilistic predictive models\nby David Widmann (Upp sala University) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nMost supervised machine learning tasks are subject to irreducible prediction errors. Probabilistic predictive models address this limitation by providing probability distributions that represent a be lief over plausible targets\, rather than point estimates. Such models can be a valuable tool in decision-making under uncertainty\, provided that t he model output is meaningful and interpretable. Calibrated models guarant ee that the probabilistic predictions are neither over- nor under-confiden t. In the machine learning literature\, different measures and statistical tests have been proposed and studied for evaluating the calibration of cl assification models. For regression problems\, however\, research has been focused on a weaker condition of calibration based on predicted quantiles for real-valued targets. In this paper\, we propose the first framework t hat unifies calibration evaluation and tests for general probabilistic pre dictive models. It applies to any such model\, including classification an d regression models of arbitrary dimension. Furthermore\, the framework ge neralizes existing measures and provides a more intuitive reformulation of a recently proposed framework for calibration in multi-class classificati on. In particular\, we reformulate and generalize the kernel calibration e rror\, its estimators\, and hypothesis tests using scalar-valued kernels\, and evaluate the calibration of real-valued regression problems.\n LOCATION:https://researchseminars.org/talk/gbgstats/7/ END:VEVENT BEGIN:VEVENT SUMMARY:Karin Hårding and Daire Carroll (University of Gothenburg) DTSTART:20221208T141500Z DTEND:20221208T150000Z DTSTAMP:20260422T122728Z UID:gbgstats/8 DESCRIPTION:Title: Population dynamics and ecology of seal populations\, empirical data and the search for theory to help our understanding. Stochastic growth models\ , image analysis\, spatial distribution and telemetry data on migrations\nby Karin Hårding and Daire Carroll (University of Gothenburg) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nT his talk is about how statistical and mathematical methods are helpful whe n we try to understand processes in wildlife populations. The European har bour seal (Sw: knubbsälen) has been studied carefully for 40 years and th e long time series allows analysis of how population growth is regulated. Recently the population growth has declined and we visited the colonies t o try to document in detail what is going on in order to give better advis e to managers. We develop new methods for estimating body size from drones and for counting seals from photos with machine learning algorithms. We a pply stochastic population growth models\, dynamic energy budget models\, and we discuss what is density dependence in age structured populations in a variable environment. We are also interested in new collaborations and feed back and look forward to interesting discussions on ways forward. Wel come! Karin and Daire\n\nKarin Harding is professor in animal ecology with a focus on marine mammals at GU and Daire Carroll is postdoctoral researc her and develops new digital tools for wildlife ecology.\n LOCATION:https://researchseminars.org/talk/gbgstats/8/ END:VEVENT BEGIN:VEVENT SUMMARY:Khalil Shafie Holighi (University of Northern Colorado) DTSTART:20221201T141500Z DTEND:20221201T150000Z DTSTAMP:20260422T122728Z UID:gbgstats/9 DESCRIPTION:Title: A test for multiple signal detection from noisy images\nby Khalil Sha fie Holighi (University of Northern Colorado) as part of Gothenburg statis tics seminar\n\nLecture held in MVL14.\n\nAbstract\nGaussian random field theory has been extensively used to model the brain images.\nIn this work \, I use the reproducing kernel Hilbert space (RKHS) machinery to derive the likelihood ratio test statistic for activation signal detection in functional magnetic resonance imaging. The models considered have the f orm of smoothed version of signal plus a white noise which include sc ale and rotation space random fields with one or more signals as special cases.\n\nKhalil Shafie Holighi is professor of statistics at University of Northern Colorado.\n LOCATION:https://researchseminars.org/talk/gbgstats/9/ END:VEVENT BEGIN:VEVENT SUMMARY:Peter Guttorp (University of Washington/Norwegian computing center ) DTSTART:20221215T141500Z DTEND:20221215T150000Z DTSTAMP:20260422T122728Z UID:gbgstats/10 DESCRIPTION:Title: Vadan och varthän?\nby Peter Guttorp (University of Washington/Norw egian computing center) as part of Gothenburg statistics seminar\n\nLectur e held in MVL14.\n\nAbstract\nSverige har haft professurer is statistik (o ch statskunskap) sedan början av 1900-talet. En av dessa var Pontus Fahlb eck i Lund\, som förespråkade enbart en samhällsvetenskaplig inriktning på statistiken. Lundaastronomen Carl Charlier sysslade med stellar stati stik\, och ansåg att statistikvetenskapen var användbar i naturvetenskap lika väl som samhällskunskap. Vi berättar hur denna konflikt resultera de i institutioner både i statistik och matematisk statistik. Vi föresl år att statistikvetenskapen\, vare sig den kallas statistik eller matemat isk statistik\, bör hamna i en egen institution.\n LOCATION:https://researchseminars.org/talk/gbgstats/10/ END:VEVENT BEGIN:VEVENT SUMMARY:Alan Gelfand (Department of Statistical Science\, Duke University) DTSTART:20230316T121500Z DTEND:20230316T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/11 DESCRIPTION:Title: Three Spatial Data Fusion Vignettes\nby Alan Gelfand (Department of Statistical Science\, Duke University) as part of Gothenburg statistics se minar\n\nLecture held in MVL14.\n\nAbstract\nWith increased collection of spatial (and spatio-temporal) datasets\, we often find multiple sources th at are capable of informing about features of a process of interest. Throu gh suitable fusion of the data sources\, we can learn at least as much abo ut the process features of interest than from any individual source. For three different illustrative ecological/environmental applications\, this talk will propose suitable coherent stochastic modeling to implement a fus ion of these sources. We focus exclusively on approaches that arise throug h generative hierarchical modeling\; the specification could produce the d ata sources that have been observed. Such modeling enables full inference both with regard to estimation and prediction\, with implicit incorporati on of uncertainty. \n\nWe consider the general setting of points and mark s\, modeled as $[points][marks|points]$\, points in $\\mathcal{D}$\, marks in $\\mathcal{Y}$. The process can model the points themselves\, the mar ks themselves (ignoring any randomness in the points)\, or the points and marks jointly. This results in four data types: (i) a point pattern\, $\\ mathcal{S}= (\\textbf{s}_{1}\, \\textbf{s}_{2}\,\\ldots\,\\textbf{s}_{n})$ \, (ii) a vector of counts for sets\, $\\{N(B_{k})\, k=1\,2\,\\ldots\,K\\} $\, (iii) a vector of observations at points\, $\\{Y(\\textbf{s}_{i})\,i=1 \,2\,\\ldots\,n\\}$\, (iv) a vector of averages for sets\, $\\{Y(B_{1})\, Y(B_{2})\,\\ldots\,Y(B_{k})\\}$. We illustrate with two data sources\; ea ch can be any one of the four data types. Regardless of how the data are observed\, we imagine the process operates at point level. Further\, we im agine a stochastic process over $\\mathcal{D}$ which links the two data so urces.\n\nThe first vignette considers presence/absence data over $\\mathc al{D}$ with one dataset being presence/absence of a species collected at a set of chosen locations. The other data source is in the form of museum/ citizen science data\, recording random locations where the species was ob served. The goal is to better understand the probability of presence surf ace over $\\mathcal{D}$. The second vignette considers zooplankton abundan ce data gathered through two different $\\it{towing}$ mechanisms. One mec hanism is calibrated while the other is not. The goal is to better unders tand zooplankton abundance over $\\mathcal{D}$. The third\, and most chal lenging vignette seeks to learn about whale abundance. Here\, the two sou rces are aerial distance sampling data for whale sightings and passive aco ustic monitoring data (using monitors on the ocean floor) for whale calls. \n\nThis is joint work with Shin Shirota\, Jorge Castillo-Mateo\, Erin Sc hliep\, and Rob Schick.\n LOCATION:https://researchseminars.org/talk/gbgstats/11/ END:VEVENT BEGIN:VEVENT SUMMARY:Martin Voigt Vejling (Aalborg University) DTSTART:20230216T121500Z DTEND:20230216T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/12 DESCRIPTION:Title: Applications of point process models to wireless communication systems\nby Martin Voigt Vejling (Aalborg University) as part of Gothenburg sta tistics seminar\n\nLecture held in MVL15.\n\nAbstract\nWireless communicat ions is a field in engineering that is attracting a lot of attention and o ne of the envisioned enablers of future communication systems is statistic al learning. Interestingly\, many scenarios considered in wireless communi cations is naturally modelled by point processes. This has historically be en studied with stochastic geometry models for wireless networks\, however \, some areas of applications remain underexplored.\n\nIn this talk\, I wi ll introduce the audience to what wireless communication is\, discussing t he basic concepts\, models\, and challenges. This includes discussing the typically used parametric channel model and how harmonic analysis\, estima tion theory\, and compressive sensing are central mathematical topics appl ied in practice. Then\, I will motivate the use of point process models wi thin wireless communications by giving a general modelling perspective. Fi nally\, I will discuss the primary focus of my research in radio frequency sensing powered by statistical learning for point processes.\n\nThe speak er is a PhD student shared between the Department of Electronic Systems an d the Department of Mathematical Sciences at Aalborg University. The talk is intended for researchers in spatial statistics interested in applicatio ns to topics in wireless communications.\n LOCATION:https://researchseminars.org/talk/gbgstats/12/ END:VEVENT BEGIN:VEVENT SUMMARY:Torgny Lindvall (Chalmers University of Technology & University of Gothenburg) DTSTART:20230202T121500Z DTEND:20230202T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/13 DESCRIPTION:Title: On coupling of renewal processes and random walks\nby Torgny Lindval l (Chalmers University of Technology & University of Gothenburg) as part o f Gothenburg statistics seminar\n\nLecture held in MVL15.\n\nAbstract\nWe use an Ornstein coupling for another proof of Blackwell's renewal theorem\ , and a Mineka coupling to establish a 0-2 law for random walks.\n LOCATION:https://researchseminars.org/talk/gbgstats/13/ END:VEVENT BEGIN:VEVENT SUMMARY:Mike Pereira (Mines Paris - PSL University) DTSTART:20230223T121500Z DTEND:20230223T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/15 DESCRIPTION:Title: Gaussian fields on Riemannian manifolds: Application to Geostatistics.\nby Mike Pereira (Mines Paris - PSL University) as part of Gothenburg s tatistics seminar\n\nLecture held in MVL15.\n\nAbstract\nMany applications in spatial statistics require data to be modeled by Gaussian processes on non-Euclidean domains\, or with non-stationary properties. Using such mo dels generally comes at the price of a drastic increase in operational cos ts (computational and storage-wise)\, rendering them hard to apply to larg e datasets. In this talk\, we propose a solution to this problem\, which r elies on the definition of a class of random fields on Riemannian manifold s. These fields extend ongoing work that has been done to leverage a chara cterization of the random fields classically used in Geostatistics as solu tions of stochastic partial differential equations. The discretization of these generalized random fields\, undertaken using a finite element approa ch\, then provides an explicit characterization that is leveraged to solve the scalability problem. Indeed\, matrix-free algorithms\, in the sense t hat they do not require to build and store any covariance (or precision) m atrix\, are derived to tackle for instance the simulation of large Gaussia n fields with given covariance properties\, even in the non-stationary set ting.\n LOCATION:https://researchseminars.org/talk/gbgstats/15/ END:VEVENT BEGIN:VEVENT SUMMARY:Konstantinos Konstantinou (Chalmers University of Technology & Uni versity of Gothenburg) DTSTART:20230309T121500Z DTEND:20230309T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/16 DESCRIPTION:Title: Global tests for quantile regression with applications in modeling distr ibutions.\nby Konstantinos Konstantinou (Chalmers University of Techno logy & University of Gothenburg) as part of Gothenburg statistics seminar\ n\nLecture held in MVL15.\n\nAbstract\nIn this talk\, I will give an intro duction to global tests for quantile regression\, i.e.\, statistical tests allowing for simultaneous inference of the quantile regression process\, with graphical interpretation. The proposed global quantile regression tes ts can determine not only if there is a difference\, but it can also deter mine for which quantiles the difference is significant on the global signi ficance level. The case where the effect of a factor (e.g.\, a categorical factor giving the group) on the distribution functions is of interest but confounded with other factors affecting the distributions is studied. An extensive simulation study is conducted to compare the global quantile reg ression tests with classical graphical tests based on the Kolmogorov-Smirn ov test statistic. This is a joint work with Tomáš Mrkvička\, Mari Myll ymäki and Mikko Kuronen.\n LOCATION:https://researchseminars.org/talk/gbgstats/16/ END:VEVENT BEGIN:VEVENT SUMMARY:Finn Lindgren (University of Edinburgh) DTSTART:20230314T121500Z DTEND:20230314T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/17 DESCRIPTION:Title: Stochastic adventures in space and time\nby Finn Lindgren (Universit y of Edinburgh) as part of Gothenburg statistics seminar\n\nLecture held i n MVL14.\n\nAbstract\nThe standard geostatistics toolbox includes methods for modelling\nspatial dependence between georeferenced observations\, as well as\nmethods for modelling the occurrence of random points. The core\ nmodel building blocks are often some form of Gaussian random fields.\n\nT he easiest approach to constructing space-time models is by taking\nthe pr oduct between a spatial covariance kernel and a temporal\ncovariance kerne l. These are called covariance separable models. An\nalternative that may better capture the spatio-temporal dynamics is to\ntake inspiration for ph ysics motivated partial differential equations\nsuch as the heat equation\ , which leads to non-separable models.\nNon-separable models are in genera l more computationally expensive\,\nbut one can sometimes use the model st ructure to retain a lot of the\nsimplicity of separable models\, for examp le allowing these models to\nbe used as components of larger hierarchical generalised additive\nmodels. For point process observations\, such as obs ervations of a\nmoving animal\, the temporal dynamics poses an additional challenge.\n\nI will discuss some of these aspects\, including a basic con struction\nof non-separable space-time models\, as well as an application of the\nINLA/inlabru framework to estimate the parameters of a dynamical\n animal movement model by rephrasing it as a point process model\, with\na parametric movement kernel\, and a random field as an unknown\n"resource s election function".\n LOCATION:https://researchseminars.org/talk/gbgstats/17/ END:VEVENT BEGIN:VEVENT SUMMARY:Nicoletta D’Angelo (University of Palermo) DTSTART:20230404T111500Z DTEND:20230404T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/18 DESCRIPTION:Title: Self-exciting point process modelling of crimes on linear networks\n by Nicoletta D’Angelo (University of Palermo) as part of Gothenburg stat istics seminar\n\nLecture held in MVL15.\n\nAbstract\nAlthough there are r ecent developments in analysing first and second-order characteristics of point processes on networks\, there are very few attempts to introduce mod els for network data.\nMotivated by the analysis of crime data in Bucarama nga (Colombia)\, we propose a spatio-temporal Hawkes point process model a dapted to events living on linear networks. We first consider a non-param etric modelling strategy\, for which we follow a non-parametric estimation of both the background and the triggering components. Then we consider a semi-parametric version\, including a parametric estimation of the backgro und based on covariates. Our network model outperforms a planar version\, improving the fitting of the self-exciting point process model\, and can b e easily adapted to multi-type processes.\n LOCATION:https://researchseminars.org/talk/gbgstats/18/ END:VEVENT BEGIN:VEVENT SUMMARY:Petar Jovanovski (Chalmers University of Technology & University o f Gothenburg) DTSTART:20230420T111500Z DTEND:20230420T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/23 DESCRIPTION:Title: Approximate Bayesian Computation with Backward Simulation for Discretely Observed Diffusions\nby Petar Jovanovski (Chalmers University of Tech nology & University of Gothenburg) as part of Gothenburg statistics semina r\n\nLecture held in MVL14.\n\nAbstract\nStochastic differential equations (SDE) are employed in many areas of science as a powerful tool for modell ing processes that are subject to random fluctuations. Bayesian inference for a large class of SDEs is challenging due to the analytic intractabilit y of the likelihood function. Nevertheless\, forward simulation via numeri cal methods is straightforward\, motivating the use of approximate Bayesia n computation (ABC). We propose a simulation scheme for SDE models that is based on processing the observation in both the forward and backward dire ction\, effectively utilizing the information provided by the observed dat a. This leads to the simulation of sample paths that are consistent with t he observations\, thereby increasing the ABC acceptance rate. We additiona lly leverage partial exchangeability of Markov processes and employ invari ant neural networks to learn the summary statistics that are needed in ABC . These are sequentially learned by exploiting a sequential Monte Carlo AB C sampler\, which provides new training data at each iteration. Therefore\ , our novel contribution is a learning tool for SDE model parameters while simultaneously learning the summary statistics. Using synthetic data gene rated from the Chan-Karaolyi-Longstaff-Sanders SDE family\, we show that o ur approach accelerates inference considerably\, compared to standard (for ward-only) methods\, while preserving inference accuracy.\n LOCATION:https://researchseminars.org/talk/gbgstats/23/ END:VEVENT BEGIN:VEVENT SUMMARY:David Bolin (King Abdullah University of Science and Technology) DTSTART:20230427T111500Z DTEND:20230427T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/24 DESCRIPTION:Title: Gaussian Whittle-Matérn fields on metric graphs\nby David Bolin (Ki ng Abdullah University of Science and Technology) as part of Gothenburg st atistics seminar\n\nLecture held in MVL14.\n\nAbstract\nWe define a new cl ass of Gaussian processes on compact metric graphs such as street or river networks. The proposed models\, the Whittle-Matérn fields\, are defined via a fractional stochastic partial differential equation on the compact m etric graph and are a natural extension of Gaussian fields with Matérn co variance functions on Euclidean domains to the non-Euclidean metric graph setting. Existence of the processes\, as well as their sample path regular ity properties are derived. The model class in particular contains differe ntiable Gaussian processes. To the best of our knowledge\, this is the fir st construction of a valid differentiable Gaussian field on general compac t metric graphs.\nWe then focus on a model subclass which we show contains processes with Markov properties. For this case\, we show how to evaluate finite dimensional distributions of the process exactly and computational ly efficiently. This facilitates using the proposed models for statistical inference without the need for any approximations. Finally\, we derive so me of the main statistical properties of the model class\, such as consist ency of maximum likelihood estimators of model parameters and asymptotic o ptimality properties of linear prediction based on the model with misspeci fied parameters. \nThe usage of the model class is illustrated through an application to traffic data.\n LOCATION:https://researchseminars.org/talk/gbgstats/24/ END:VEVENT BEGIN:VEVENT SUMMARY:Moa Johansson (Chalmers University of Technology) DTSTART:20230504T111500Z DTEND:20230504T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/25 DESCRIPTION:Title: Machine Learning Methods for texts from Political Science\nby Moa Jo hansson (Chalmers University of Technology) as part of Gothenburg statisti cs seminar\n\nLecture held in MVL14.\n\nAbstract\nIn the WASP-HS project " Bias and Methods of AI Technology Studying Political Behavior" we are inve stigating and developing machine learning methods to help political scient ists study the enormous amounts of text documents that are otherwise beyon d manual analysis\, such as the document repository from the Swedish Riksd ag. This is a collaborative project between Dr. Annika Fredén's group in the Political Science department at Lund University\, and the group of Dr. Moa Johansson at Computer Science at Chalmers.\n\nI will give an overview of some of the work so far on how we are trying to highlight differences in language use between parties in the Swedish Riksdag. The first paper is about comparing word embeddings trained on texts from different parties. The second concerns explainability of text classification: if a machine le arning algorithm can classify text as belonging to one party or another\, it is useful for a social scientist to know what such a classification is based on. We have started to develop a new method for class explainability for text for this purpose. This is going work with PhD student Denitsa Sa ynova\, and post-doc Bastiaan Bruinsma.\n LOCATION:https://researchseminars.org/talk/gbgstats/25/ END:VEVENT BEGIN:VEVENT SUMMARY:Oskar Allerbo (Chalmers University of Technology & University of G othenburg) DTSTART:20230511T111500Z DTEND:20230511T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/26 DESCRIPTION:Title: Solving Kernel Ridge Regression with Gradient Descent\nby Oskar Alle rbo (Chalmers University of Technology & University of Gothenburg) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nW e present an equivalent formulation for the objective function of kernel r idge regression (KRR)\, that opens up for studying KRR from the perspectiv e of gradient descent. Utilizing gradient descent with infinitesimal step size\, allows us to formulate a new regularization for kernel regression through early stopping.\n\nThe gradient descent formulation of KRR allo ws us expand to a time dependent stationary kernel\, where we decrease the bandwidth to zero during training. This circumvents the need of hyper par ameter selection. Furthermore\, we are able to achieve both zero train ing error and a double descent behavior\, phenomena that do not occur for KRR with constant bandwidth\, but are known to appear for neural networks. \n\nThe new formulation of KRR also enables us to explore other penalties than the ridge penalty. Specifically\, we explore the $\\ell_1$ and $\\ell _\\infty$ penalties and show that these correspond to two flavors of gradi ent descent\, thus alleviating the need of computationally heavy proximal gradient descent algorithms. We show theoretically and empirically how the se formulations correspond to signal-driven and robust regression\, respec tively.\n LOCATION:https://researchseminars.org/talk/gbgstats/26/ END:VEVENT BEGIN:VEVENT SUMMARY:Juan Inda (Chalmers University of Technology & University of Gothe nburg) DTSTART:20230516T111500Z DTEND:20230516T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/27 DESCRIPTION:Title: Confidence-based Prediction of Antibiotic Resistance at the Patient-leve l Using Transformers\nby Juan Inda (Chalmers University of Technology & University of Gothenburg) as part of Gothenburg statistics seminar\n\nLe cture held in MVL15.\n\nAbstract\nRapid and accurate diagnostics of bacter ial infections are necessary for efficient treatment of antibiotic-resista nt pathogens. Cultivation-based methods\, such as antibiotic susceptibilit y testing (AST)\, are slow\, resource-demanding\, and can fail to produce results before the treatment needs to start. This increases patient risks and antibiotic overprescription. Here\, we present a deep-learning method that uses transformers to merge patient data with available AST results to predict antibiotic susceptibilities that have not been measured. The meth od is combined with conformal prediction (CP) to enable the estimation of uncertainty at the patient-level. After training on three million AST resu lts from thirty European countries\, the method made accurate predictions for most antibiotics while controlling the error rates\, even when limited diagnostic information was available. We conclude that transformers and C P enables confidence-based decision support for bacterial infections and\, thereby\, offer new means to meet the growing burden of antibiotic resist ance.\n LOCATION:https://researchseminars.org/talk/gbgstats/27/ END:VEVENT BEGIN:VEVENT SUMMARY:Frank Miller (Linköping University and Stockholm University) DTSTART:20230525T111500Z DTEND:20230525T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/28 DESCRIPTION:Title: Parallel optimal pretesting of mixed-format questions for achievement te sts\nby Frank Miller (Linköping University and Stockholm University) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbst ract\nWhen large achievement tests like national tests in school or admiss ion tests for university are conducted regularly\, the test-questions need to be pretested before being used in the test. Methods for assigning pret esting questions to examinees in an optimal way based on their ability hav e been developed. Most of these methods are intended for a situation where examinees arrive sequentially for being assigned to pretesting questions. However\, several pretests (e.g.\, for national tests in Swedish schools or for högskoleprovet) are conducted in a way where all or many examinees conduct the test in parallel. In this talk\, we develop an optimal design for such parallel pretest setups which can be implemented in real scenari os. In many real test situations\, questions are of mixed format and our o ptimal design method can handle that. We discuss first the optimal designs for the 2-parameter logistic\, the 3-parameter logistic\, and the general ized partial credit model. Then\, we consider the case of mixed-format tes ts where all these models are used to fit the data. The method we propose can also take different expected solve times into consideration. We invest igate the efficiency gain of the method. Our investigations show that the proposed method is able to increase the efficiency of pretests considerabl y. The described method has been used for the Swedish national tests in ma thematics. \n\nThis is a joint work with Ellinor Fackle-Fornius\n LOCATION:https://researchseminars.org/talk/gbgstats/28/ END:VEVENT BEGIN:VEVENT SUMMARY:Lotta Eriksson (Chalmers University of Technology & University of Gothenburg) DTSTART:20230601T111500Z DTEND:20230601T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/29 DESCRIPTION:Title: A multitype Galton-Watson model of biological aging\nby Lotta Erikss on (Chalmers University of Technology & University of Gothenburg) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nTh e progressive accumulation of damaged proteins within the cellular structu re is a key factor in the aging process of yeast. By considering the quant ity of damaged proteins as a measure of the cell's biological age\, we exp lore an individual-based stochastic population model that incorporates rej uvenation events. During cell division\, the mother cell is given the oppo rtunity rejuvenate\, by transferring the accumulated damage to the daughte r cell. This modeling approach allows us to study the dynamics of the agin g process and understand the impact of rejuvenation on the overall populat ion.\n LOCATION:https://researchseminars.org/talk/gbgstats/29/ END:VEVENT BEGIN:VEVENT SUMMARY:Felix Held (Chalmers University of Technology & University of Goth enburg) DTSTART:20230608T111500Z DTEND:20230608T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/30 DESCRIPTION:Title: Simultaneous gene clustering and regulatory program reconstruction revea ls insight into the phenotypic plasticity of neural cancers\nby Felix Held (Chalmers University of Technology & University of Gothenburg) as par t of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\n Nervous system cancers contain a large spectrum of transcriptional cell st ates\, reflecting processes active during normal development\, injury resp onse and growth. However\, we lack a good understanding of these states' r egulation and pharmacological importance. Here\, we describe the integrate d reconstruction of such cellular regulatory programs and their therapeuti c targets from extensive collections of single-cell RNA sequencing data (s cRNA-seq). Our approach called single-cell Regulatory-driven Clustering (s cRegClust) performs simultaneous gene clustering and regulatory program re construction tasks to predict essential kinases and transcription factors. We formulate an apriori intractable partitioning problem that connects ge ne modules with linear regulator models. A greedy two-step procedure is co nstructed to iteratively update gene modules and associated regulatory pro grams and find an approximate solution. Penalized regression was used to r eplace a combinatorial selection problem in the construction of regulatory programs and predictive modelling was used during gene cluster allocation . The method is used to identify regulatory programs in tumor cell states from both adult and childhood brain cancers. Further analysis corroborated by experimental results leads to hypothesis generation on an underlying b iological mechanism for drug combination therapy in adult glioblastoma.\n LOCATION:https://researchseminars.org/talk/gbgstats/30/ END:VEVENT BEGIN:VEVENT SUMMARY:Sara Hamis (Tampere University) DTSTART:20231026T111500Z DTEND:20231026T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/31 DESCRIPTION:Title: Spatial cumulant models for mathematical cancer research\nby Sara Ha mis (Tampere University) as part of Gothenburg statistics seminar\n\nLectu re held in MVL14.\n\nAbstract\nSpatial cumulant models (SCMs) are spatiall y resolved population models\, formulated by differential equations\, that describe population dynamics generated by spatio-temporal point processes (STPPs). Specifically\, SCMs approximate the dynamics of two STPP-generat ed summary statistics: first-order spatial cumulants (densities)\, and sec ond-order spatial cumulants (spatial covariances). \n\nIn this talk\, I’ ll exemplify how SCMs can be used in mathematical oncology by modelling th eoretical cancer cell populations comprising interacting subclones. Our re sults demonstrate that SCMs can capture STPP-generated population density dynamics\, even when mean-field population models (MFPMs) fail to do so. F rom both MFPM and SCM equations\, we derive treatment-induced death rates required to achieve non-growing cell populations. When testing these treat ment strategies in STPP-generated cell populations\, our results demonstra te that SCM-informed strategies outperform MFPM-informed strategies in ter ms of inhibiting population growths. We thus demonstrate that SCMs provide a new framework in which to study cell-cell interactions and treatments t hat take cell-cell interactions into account. \n\nJoint work with: Panu So mervuo\; J. Arvid Ågren\; Dagim S. Tadele\; Juha Kesseli\; Jacob G. Scott \; Matti Nykter\; Philip Gerlee\; Dmitri Finkelshtein\; Otso Ovaskainen.\n LOCATION:https://researchseminars.org/talk/gbgstats/31/ END:VEVENT BEGIN:VEVENT SUMMARY:Hrvoje Planinić (University of Zagreb) DTSTART:20230921T091500Z DTEND:20230921T100000Z DTSTAMP:20260422T122728Z UID:gbgstats/32 DESCRIPTION:Title: Extremes of stationary heavy-tailed time series\nby Hrvoje Planinić (University of Zagreb) as part of Gothenburg statistics seminar\n\nLectur e held in MVL14.\n\nAbstract\nWe will present a framework for describing t he asymptotic behavior of high-level exceedances for stationary (i.e. depe ndent) time series with heavy-tailed marginal distribution and whose excee dances occur in clusters\; think of modelling e.g. financial returns or da ily rainfall measurements. The main tools are the theory of point processe s and the notion of the so-called tail process. The latter allows one to f ully describe the asymptotic distribution of the extremal clusters using t he language of standard Palm theory. We will illustrate the general theory on simple moving average models. If time permits\, we will comment on how this framework can be extended to deal with extremes related to models fr om stochastic geometry.\n LOCATION:https://researchseminars.org/talk/gbgstats/32/ END:VEVENT BEGIN:VEVENT SUMMARY:Devdatt Dubhashi (Chalmers) DTSTART:20230928T111500Z DTEND:20230928T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/33 DESCRIPTION:Title: Bandits: Structured and Constrained\nby Devdatt Dubhashi (Chalmers) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbst ract\nI will give an introduction and invitation to Bandits - a very simpl e\, yet central model of sequential decision making under uncertainty. Aft er introducing the central concepts and some of the basic algorithms and r esults\, I'll describe some recent work from our group on some extensions of the basic model which are also useful in applications. Throughout I'll try to show how the subject has close connections to information theory\, statistics and optimization.\n LOCATION:https://researchseminars.org/talk/gbgstats/33/ END:VEVENT BEGIN:VEVENT SUMMARY:Serik Sagitov (Chalmers University of Technology & university of G othenburg) DTSTART:20231103T121500Z DTEND:20231103T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/35 DESCRIPTION:Title: Theta-positive branching processes in varying environment\nby Serik Sagitov (Chalmers University of Technology & university of Gothenburg) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstrac t\nBranching processes in a varying environment encompass a wide range of stochastic demographic models\, and their complete understanding in terms of limit behavior poses a formidable research challenge. In this paper\, w e conduct a thorough investigation of such processes within a continuous-t ime framework\, assuming that the reproduction law of individuals adheres to a specific parametric form for the probability generating function. Our six clear-cut limit theorems support the notion of recognizing five disti nct asymptotical regimes for branching in varying environments: supercriti cal\, asymptotically degenerate\, critical\, strictly subcritical\, and lo osely subcritical.\n LOCATION:https://researchseminars.org/talk/gbgstats/35/ END:VEVENT BEGIN:VEVENT SUMMARY:Olle Häggström (Chalmers University of Technology & university o f Gothenburg) DTSTART:20231006T111500Z DTEND:20231006T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/36 DESCRIPTION:Title: Playing with fire\nby Olle Häggström (Chalmers University of Techn ology & university of Gothenburg) as part of Gothenburg statistics seminar \n\nLecture held in MVL14.\n\nAbstract\nAlan Turing speculated in 1951 abo ut a time point in the future when machines “outstrip our feeble powers ” in such a way that we lose our position as the most intelligent specie s on the planet. Current AI trends suggest that we are rapidly approaching that time point. This is playing with fire\, because at such a time point our continued wellbeing hinges largely on what the first superintelligent machines are motivated to do. If their goals and values are aligned with ours\, then a brilliant future awaits us\, while if not\, then most likely it is (in the words of OpenAI’s CEO Sam Altman) “lights out for every one”. Making this transition go well involves considerable technological and societal challenges.\n LOCATION:https://researchseminars.org/talk/gbgstats/36/ END:VEVENT BEGIN:VEVENT SUMMARY:Moritz Schauer (Chalmers University of Technology & University of Gothenburg) DTSTART:20231020T090000Z DTEND:20231020T094500Z DTSTAMP:20260422T122728Z UID:gbgstats/37 DESCRIPTION:Title: Causal structure learning and sampling using Markov Monte Carlo with mom entum\nby Moritz Schauer (Chalmers University of Technology & Universi ty of Gothenburg) as part of Gothenburg statistics seminar\n\n\nAbstract\n In the context of inferring a Bayesian network structure from observationa l data\, that is inferring a directed acyclic graph (DAG)\, we devise a no n-reversible continuous-time Markov chain that targets a probability distr ibution over classes of observationally equivalent (Markov equivalent) DAG s. The classes are represented as completed partially directed acyclic gra phs (CPDAGs). The non-reversible Markov chain relies on the operators used in Chickering’s Greedy Equivalence Search (GES) and is endowed with a m omentum variable\, which improves mixing significantly as we show empirica lly. The possible target distributions include posterior distributions bas ed on a prior and a Markov equivalent likelihood. Joint work with Marcel W ienöbst (Universität zu Lübeck).\n\nThis is a talk in the webinar serie s of the Cramér society\n LOCATION:https://researchseminars.org/talk/gbgstats/37/ END:VEVENT BEGIN:VEVENT SUMMARY:Frédéric Lavancier (Nantes University\, France) DTSTART:20240119T121500Z DTEND:20240119T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/38 DESCRIPTION:Title: Spatial birth-death-move processes: basic properties and inference\n by Frédéric Lavancier (Nantes University\, France) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nBirth-death-mov e processes are Markov models for the spatio-temporal dynamics of a system of particles in motion where births and deaths can occur. Natural applica tions include epidemiology\, individual-based modelling in ecology and spa tio-temporal dynamics observed in bio-imaging. We present some of the basi c probabilistic properties of these processes and we consider two inferenc e problems: First\, the non-parametric estimation of the birth and death i ntensity functions\; Second\, the parametric estimation of the full dynami cs by maximum likelihood. We finally apply our statistical method to the a nalysis of a real dataset representing the spatio-temporel dynamics of bio molecules observed in a living cell.\n LOCATION:https://researchseminars.org/talk/gbgstats/38/ END:VEVENT BEGIN:VEVENT SUMMARY:Christian Hirsch (Aarhus University) DTSTART:20240301T121500Z DTEND:20240301T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/41 DESCRIPTION:Title: On the topology of higher-order age-dependent random connection models\nby Christian Hirsch (Aarhus University) as part of Gothenburg statisti cs seminar\n\nLecture held in MVL14.\n\nAbstract\nPreferential attachment is a popular mechanism for generating scale-free networks. While it offers a compelling narrative\, the underlying reinforced processes make it diff icult to rigorously establish subtle properties. Recently\, age-dependent random connection models were proposed as an alternative that is capable o f generating similar networks with a mechanism that is amenable to a more refined analysis. In this talk\, we analyze the asymptotic behavior of hig her-order topological characteristics such as higher-order degree distribu tions and Betti numbers in large domains. We demonstrate the practical ap plication of the theoretical results to real-world datasets by analyzing s cientific collaboration networks based on data from arXiv.This talk is bas ed on joint work with Péter Juhász\n LOCATION:https://researchseminars.org/talk/gbgstats/41/ END:VEVENT BEGIN:VEVENT SUMMARY:Pierre Nyquist (Chalmers University of Technology & University of Gothenburg) DTSTART:20240221T121500Z DTEND:20240221T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/42 DESCRIPTION:Title: Large deviations for Markov chain Monte Carlo methods: the surprisingly curious case of Metropolis-Hastings.\nby Pierre Nyquist (Chalmers Univ ersity of Technology & University of Gothenburg) as part of Gothenburg sta tistics seminar\n\nLecture held in MVL14.\n\nAbstract\nMarkov chain Monte Carlo (MCMC) methods have become the workhorse for numerical computations in a range of scientific disciplines\, e.g.\, computational chemistry and physics\, statistics\, and machine learning. The performance of MCMC metho ds has therefore become an important topic at the intersection of probabil ity theory and (computational) statistics: e.g.\, when the underlying dist ribution one is trying to sample from becomes sufficiently complex\, conve rgence speed and/or the cost per iteration becomes an issue for most MCMC methods. \n\nThe analysis\, and subsequently design\, of MCMC methods has to a large degree relied on classical tools used to determine the speed of convergence of Markov chains\, e.g.\, mixing times\, spectral gap and fun ctional inequalities (Poincaré\, log-Sobolev). An alternative avenue is t o use the theory of large deviations for empirical measures. In this talk I will first give a general outline of this approach to analysing MCMC met hods\, along with some recent examples. I will then consider the specific case of the Metropolis-Hastings algorithm\, the most classical amongst all MCMC methods and a foundational building block for many more advanced met hods. Despite the simplicity of this method\, it turns out that the theore tical analysis of it is still a rich area\, and from the large deviation p erspective it is surprisingly difficult to treat. As a first step we show a large deviation principle for the underlying Markov chain\, extending th e celebrated Donsker-Varadhan theory. Time permitted I will also discuss o ngoing and future work on using this result for better understanding of bo th the Metropolis-Hastings method and more advanced methods\, such as appr oximate Bayesian computation (ABC-MCMC) and the Metropolis-adjusted Langev in algorithm (MALA).\n\nThe talk will be self-contained and no prior knowl edge of either MCMC methods or large deviations is required.\n\nThe talk i s primarily based on join work with Federica Milinanni (KTH).\n LOCATION:https://researchseminars.org/talk/gbgstats/42/ END:VEVENT BEGIN:VEVENT SUMMARY:Vincent Szolnoky (Chalmers University of Technology & University o f Gothenburg) DTSTART:20240306T121500Z DTEND:20240306T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/45 DESCRIPTION:Title: On the Interpretability of Regularisation for Neural Networks Through Mo del Gradient Similarity\nby Vincent Szolnoky (Chalmers University of T echnology & University of Gothenburg) as part of Gothenburg statistics sem inar\n\nLecture held in MVL14.\n\nAbstract\nMost complex machine learning and modelling techniques are prone to over-fitting and may subsequently ge neralise poorly to future data. Artificial neural networks are no differen t in this regard and\, despite having a level of implicit regularisation w hen trained with gradient descent\, often require the aid of explicit regu larisers. We introduce a new framework\, Model Gradient Similarity (MGS)\, that (1) serves as a metric of regularisation\, which can be used to moni tor neural network training\, (2) adds insight into how explicit regularis ers\, while derived from widely different principles\, operate via the sam e mechanism underneath by increasing MGS\, and (3) provides the basis for a new regularisation scheme which exhibits excellent performance\, especia lly in challenging settings such as high levels of label noise or limited sample sizes.\n LOCATION:https://researchseminars.org/talk/gbgstats/45/ END:VEVENT BEGIN:VEVENT SUMMARY:Selma Tabakovic (Chalmers University of Technology & University of Gothenburg) DTSTART:20240313T121500Z DTEND:20240313T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/46 DESCRIPTION:Title: AI-driven sepsis care: early detection and personalized treatment\nb y Selma Tabakovic (Chalmers University of Technology & University of Gothe nburg) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\ n\nAbstract\nSepsis is a life-threatening organ dysfunction caused by a dy sregulated host response to infection\, and remains a leading cause of dea th in intensive care units worldwide. An optimal treatment strategy is sti ll unknown\, leading to a significant variability in sepsis treatment with poorer outcomes.\n\nRecently\, deep reinforcement learning has shown prom ise as a decision-aiding tool for the administration of intravenous fluids and vasopressors to septic patients. However\, these models are limited i n their ability to accommodate the entire range from high-risk to low-risk patients\, and thus fail to provide personalized treatment recommendation s.\n\nTo address this limitation\, in particular in the presence of hetero geneous patient groups or heterogeneous treatment responses\, we propose a Multi-Head Dueling Double Deep Q-Network (MH-DQN) model that incorporates patient characteristics to enable more personalized treatment recommendat ions. The MH-DQN model has multiple output layers\, each of which is optim ized for a specific patient profile. The model is trained using the Medica l Information Mart for Intensive Care (MIMIC-III) database.\n LOCATION:https://researchseminars.org/talk/gbgstats/46/ END:VEVENT BEGIN:VEVENT SUMMARY:Akash Sharma (Chalmers University of Technology & University of Go thenburg) DTSTART:20240320T121500Z DTEND:20240320T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/47 DESCRIPTION:Title: Sampling on manifolds via Langevin diffusion\nby Akash Sharma (Chalm ers University of Technology & University of Gothenburg) as part of Gothen burg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nWe derive e rror bounds for sampling and estimation using a discretization of an intri nsically defined Langevin diffusion on a compact Riemannian manifold. Two estimators of linear functionals of invariant measure based on the discret ized Markov process are considered: a time-averaging estimator and an ense mble-averaging estimator. Imposing no restrictions beyond a nominal level of smoothness on potential function\, first-order error bounds\, in discre tization step size\, on the bias and variances of both estimators are deri ved. We will also discuss conditions for extending analysis to the case of non-compact manifolds and different variants of the algorithm. We will pr esent numerical illustrations with distributions on the manifolds of posit ive and negative curvature which verify the derived bounds.\n\nJoint work with Karthik Bharath (University of Nottingham)\, Alexander Lewis (Univers ity of Gottingen) and Michael Tretyakov (University of Nottingham)\n LOCATION:https://researchseminars.org/talk/gbgstats/47/ END:VEVENT BEGIN:VEVENT SUMMARY:Nathan Gillot (University of Lorraine) DTSTART:20240403T111500Z DTEND:20240403T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/48 DESCRIPTION:Title: Modelling of the spread of a pathogen in coniferous forests and modellin g for cosmological data characterization\nby Nathan Gillot (University of Lorraine) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nAs stated in the title\, this presentation will be div ided into two parts. The first will deal with work on epidemiological data in a coniferous forest. We carried out modeling by considering probabilit y laws on a lattice\, used the Gibbs sampler for simulation and used two p arameter approximation methods for these models: pseudo-likelihood and a B ayesian method\, the ABC Shadow algorithm. The second part of the talk wil l focus on cosmological data. This time\, modeling will be done by spatial point processes\, simulation by the Metropolis Hastings algorithm and inf erence again by the ABC Shadow algorithm. The aim of this presentation is to give an idea of the tools used for the modelling\, simulation and infer ence process for these two projects.\n LOCATION:https://researchseminars.org/talk/gbgstats/48/ END:VEVENT BEGIN:VEVENT SUMMARY:Sergei Zuyev (Chalmers) DTSTART:20240417T111500Z DTEND:20240417T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/49 DESCRIPTION:Title: Training Bayesian neural networks with measure optimisation algorithms\nby Sergei Zuyev (Chalmers) as part of Gothenburg statistics seminar\n\ nLecture held in MVL14.\n\nAbstract\nOn a high abstraction level\, a Bayes ian neural network (BNN) can be seen as a function\nof input data and thei r prior probability distribution which yields\,\namong other outputs\, the ir estimated posterior probability\ndistribution. This distribution is a r esult of optimisation of a\nchosen score function aiming to favour these p robability distributions\nwhich describe best the observed data and take i nto account the prior\ndistribution.\n\nInstead of constraint optimisation over the simplex of probability distributions\, it is typical to\nmap thi s simplex into Euclidean space\, for example with Softmax function or its variants\, and then do optimisation in the whole\nspace without constrain ts. It is\, however\, widely acknowledged that such mapping often suffers from undesirable properties for optimisation and\nstability of the algorit hms. To counterfeit this\, a few regularisation procedures have been propo sed in the literature.\n\nInstead of trying to modify the mapping approac h\, we suggest\nturning back to optimisation on the original simplex using recently\ndeveloped algorithms for constrained optimisation of functional s of measures. \nWe demonstrate that our algorithms run tens times faster\ nthan the standard algorithms involving softmax mapping and lead to exact solutions rather than to their approximations.\n LOCATION:https://researchseminars.org/talk/gbgstats/49/ END:VEVENT BEGIN:VEVENT SUMMARY:Radu Stoica (Université de Lorraine) DTSTART:20240424T111500Z DTEND:20240424T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/51 DESCRIPTION:Title: Approximated inference for marked Gibbs point process\nby Radu Stoic a (Université de Lorraine) as part of Gothenburg statistics seminar\n\nLe cture held in MVL14.\n\nAbstract\nParameter estimation for point processes is achieved via solving optimisation problems built using general strateg ies. Three well established strategies are enumerated. The first consists of considering contrast fuctions based on summary statistics. The second o ne uses the pseudo-likelihood. And the third approximates the likelihood f unction via Monte Carlo procedures. Each of these techniques has known adv antages and drawbacks (Moler and Waagepetersen 2004\, van Lieshout 2001\, 2019).\n\nSampling point process posterior densities is an inference appro ach deeply intertwinned wih the previous one\, since it allows simultaneou s parameter estimation and statistical tests based on observations. The au xiliary variable method (Moller et al.\,2006) gives the mathematical solut ion to this problem\, while pointing out the difficulties of its practical implementation due to poor mixing. The exchange algorithm proposed by (Mu rray et al. 2006)\, (Caimo and Friel\, 2011) proposes a solution for the p oor mixing induced by the auxiliary variable method. As its predecessor it requires exact simulation for the sampling of the auxiliary variable. Thi s is not really a drawback\, but it may explode the computational time for models exhibiting strong interactions (van Lieshout and Stoica\, 2006). \ n\nThis talk presents the approximate ABC Shadow and SSA methods as comple mentary inference methods to the ones based on posterior density sampling. These methods do not require exact simulation\, while providing the neces sary theoretical control. The derived algorithms are applied on data from several application domains such as astronomy\, geosciences and network s ciences (Stoica et al.\,17)\, (Stoica et al.\,21)\, (Hurtado et al.\,21)\, (Laporte et al.\,22).\n LOCATION:https://researchseminars.org/talk/gbgstats/51/ END:VEVENT BEGIN:VEVENT SUMMARY:Robert Berman (Chalmers University of Technology & University of G othenburg) DTSTART:20240508T111500Z DTEND:20240508T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/53 DESCRIPTION:Title: Emergent complex geometry\nby Robert Berman (Chalmers University of Technology & University of Gothenburg) as part of Gothenburg statistics se minar\n\nLecture held in MVL14.\n\nAbstract\nA recurrent theme in geometry is the quest for canonical metrics on a given manifold X. The prototypica l case is when X is a compact orientable two-dimensional surface. Such a m anifold can be endowed with a metric of constant curvature\, which is uniq uely determined by a fixing a complex structure on X. However\, from a phy sical point of view\, geometrical shapes - as we know them from everyday e xperience - are\, of course\, not fundamental physical entities. They mere ly arise as macroscopic emergent features of ensembles of microscopic poin t particles in the limit as the number N of particles tends to infinity. T his leads one to wonder if there is a canonical random point process on a given complex manifold X\, from which a canonical metrics emerges as the n umber N of points tends to infinity? This is\, indeed\, the case\, when X is a complex algebraic hypersurface of any dimension\, as explained in the present talk. In this case the emerging metrics in question have constant Ricci curvature. More precisely\, they are Kähler-Einstein metrics. The talk is aimed to be non-technical and no previous background in complex ge ometry is required.\n LOCATION:https://researchseminars.org/talk/gbgstats/53/ END:VEVENT BEGIN:VEVENT SUMMARY:Umberto Picchini (Chalmers University of Technology & University o f Gothenburg) DTSTART:20240515T111500Z DTEND:20240515T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/54 DESCRIPTION:Title: Fast\, lightweight and semi-amortised simulation-based inference\nby Umberto Picchini (Chalmers University of Technology & University of Gothe nburg) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\ n\nAbstract\nBayesian inference for complex models with an intractable lik elihood can be tackled using algorithms performing many calls to computer simulators. These approaches are collectively known as "simulation-based i nference" (SBI). Recent SBI methods use neural-conditional-estimation\, th at is neural networks are employed to provide approximations to the likeli hood function or the posterior distribution of model parameters. While neu ral-based posterior and likelihood estimation methods have produced except ionally flexible inference strategies\, these can be computationally inten sive to run and have a non-negligible impact on energy expenditure and mem ory requirements. In this work\, rather than using neural networks we prop ose more "frugal" strategies that display state-of-art inference quality\, while being able to run with limited resources\, being much faster to tra in and exhibiting a much smaller computational footprint. We investigate s tructured mixtures of probability distributions and design a new SBI metho d named Sequential Mixture Posterior and Likelihood Estimation (SeMPLE). S eMPLE learns closed-form approximations for both the posterior $p(θ|y)$ a nd the likelihood $p(y|θ)$ from the same training data\, using Gaussian m ixture models that can be efficiently learned.\nWe show favorable results for a variety of stochastic models (including SDEs and Markov jump process es)\, also in presence of multimodal posteriors. \n\nThe talk will be appr oachable for the uninitiated audience\, while novel results will be of int erest for the experienced audience.\n\nJoint work with Henrik Häggström\ , Pedro L. C. Rodrigues\, Geoffroy Oudoumanessah and Florence Forbes\, htt ps://arxiv.org/abs/2403.07454\n LOCATION:https://researchseminars.org/talk/gbgstats/54/ END:VEVENT BEGIN:VEVENT SUMMARY:Adrien Corenflos (University of Warwick) DTSTART:20240529T111500Z DTEND:20240529T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/57 DESCRIPTION:Title: Particle-MALA and Particle-mGrad: Gradient-based MCMC methods for high-d imensional state-space models\nby Adrien Corenflos (University of Warw ick) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\ nAbstract\nState-of-the-art methods for Bayesian inference in state-space models are (a) conditional sequential Monte Carlo (CSMC) algorithms\; (b) sophisticated 'classical' MCMC algorithms like MALA\, or mGRAD from Titsia s and Papaspiliopoulos (2018). The former propose N particles at each time step to exploit the model's 'decorrelation-over-time' property and thus s cale favourably with the time horizon\, T\, but break down if the dimensio n of the latent states\, D\, is large. The latter leverage gradient/prior- informed local proposals to scale favourably with D but exhibit sub-optima l scalability with T due to a lack of model-structure exploitation. We int roduce methods which combine the strengths of both approaches. The first\, Particle-MALA\, spreads N particles locally around the current state usin g gradient information\, thus extending MALA to T>1 time steps and N>1 pro posals. The second\, Particle-mGRAD\, additionally incorporates (condition ally) Gaussian prior dynamics into the proposal\, thus extending the mGRAD algorithm. We prove that Particle-mGRAD interpolates between CSMC and Par ticle-MALA\, resolving the 'tuning problem' of choosing between CSMC (supe rior for highly informative prior dynamics) and Particle-MALA (superior fo r weakly informative prior dynamics). We similarly extend other 'classical ' MCMC approaches like auxiliary MALA\, aGRAD\, and preconditioned Crank-N icolson-Langevin (PCNL). In experiments\, our methods substantially improv e upon both CSMC and sophisticated `classical' MCMC approaches for both hi ghly and weakly informative prior dynamics.\n\nTL\;DR: We aim to solve the curse of dimensionality in state-space model inferences by combining the nice property (in time) of conditional particle filtering methods\, with t he nice property (in space) of MALA and other gradient-based algorithms.\n LOCATION:https://researchseminars.org/talk/gbgstats/57/ END:VEVENT BEGIN:VEVENT SUMMARY:Helga Olafsdottir (Chalmers University of Technology & University of Gothenburg) DTSTART:20240821T111500Z DTEND:20240821T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/62 DESCRIPTION:Title: Scoring rule inference for spatial statistics based on cross-validation< /a>\nby Helga Olafsdottir (Chalmers University of Technology & University of Gothenburg) as part of Gothenburg statistics seminar\n\nLecture held in MVF21 (sic!).\n\nAbstract\nAlthough scoring rules are traditionally aimed at model evaluation\, they have also successfully been used for model inf erence. We propose parameter inference of spatial models through a leave-o ne-out cross-validation approach (LOOS)\, where the predictive ability is optimised instead of the likelihood. The approach is studied for different Gaussian spatial models. For Gaussian models with sparse precision matric es\, such as spatial Markov models\, the approach results in fast computat ions compared to the likelihood approach. Moreover\, the approach allows a ffecting the robustness to outliers and sensitivity to non-stationarity. A pplying the LOOS to ERA5 temperature reanalysis data for the contiguous Un ited States and the average July temperature for the years 1940 to 2023 re sulted in estimates with better predictive performance than the maximum li kelihood in a fraction of the computation time.\n LOCATION:https://researchseminars.org/talk/gbgstats/62/ END:VEVENT BEGIN:VEVENT SUMMARY:Erik Jansson (Chalmers University of Technology and University of Gothenburg) DTSTART:20240828T111500Z DTEND:20240828T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/63 DESCRIPTION:Title: Sampling non-stationary Gaussian random fields on hypersurfaces using su rface finite element methods\nby Erik Jansson (Chalmers University of Technology and University of Gothenburg) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nIn the middle of the previo us century\, Peter Whittle demonstrated that Whittle–Matérn Gaussian\nr andom fields on Euclidean domains can be obtained as solutions to fraction al elliptic stochastic\npartial differential equations (SPDEs). The SPDE –random field connection can be leveraged to\ngenerate random fields on other domains\, such as curves or surfaces\, by solving an SPDE on\nthat d omain. Selecting a differential operator with variable coefficients\, we o btain a flexible\nclass of non-stationary random fields. We consider how t he computational technique of surface\nfinite elements can be utilized to sample random fields on surfaces and briefly discuss how\nstrong error bou nds are obtained using complex analysis and operator theory. \nThis talk is based on joint work with Annika Lang and Mike Pereira.\n LOCATION:https://researchseminars.org/talk/gbgstats/63/ END:VEVENT BEGIN:VEVENT SUMMARY:Peter Guttorp (Norsk Regnesentral) DTSTART:20241113T121500Z DTEND:20241113T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/64 DESCRIPTION:Title: Water is warmer than air\, so why do we use sea surface temperature to e stimate global temperature?\nby Peter Guttorp (Norsk Regnesentral) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstrac t\nIn the study of global climate\, ocean temperature estimates use sea su rface temperature (SST) anomalies instead of marine atmospheric temperatur e (MAT) anomalies. A key question is to ask what biases result from this c hoice. In this talk we employ hierarchical statistical models to investiga te spatial-temporal differences between SST and MAT anomalies in the tropi cal Pacific. The analysis uses observations from the Tropical Atmosphere O cean (TAO) buoy network. We use a spatio-temporal modeling approach that a ccounts for missing data in the observation network\, and allows for full uncertainty quantification. We also compare our results to another analys is that uses data from the European Center for Medium Range Weather Foreca sting fifth generation reanalysis product (ERA5). We find no indication of bias or trend in replaciing MAT by SST in calculating global temperature anomalies.\n LOCATION:https://researchseminars.org/talk/gbgstats/64/ END:VEVENT BEGIN:VEVENT SUMMARY:Satish Iyengar (Department of Statistics\, University of Pittsburg h\, Pittsburgh\, PA\, USA) DTSTART:20240925T111500Z DTEND:20240925T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/65 DESCRIPTION:Title: A clustering problem arising in psychiatry\nby Satish Iyengar (Depar tment of Statistics\, University of Pittsburgh\, Pittsburgh\, PA\, USA) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstra ct\nCurrent psychiatric diagnoses are based primarily on self-reported exp eriences. Unfortunately\, treatments for the diagnoses are not effective f or all patients. One hypothesized reason is that ``artificial grouping of heterogeneous syndromes with different pathophysiological mechanisms into one disorder.'' To address this problem\, the US National Institute of Men tal Health instituted the Research Domain Criteria framework in 2009. This research framework calls for integrating data from many levels of informa tion: genes\, cells\, molecules\, circuits\, physiology\, behavior\, and s elf-report. Clustering comes to the forefront as a key tool in this effort . In this talk\, I present a case study of the use of mixture models to cl uster older adults based on measures of sleep from three domains: diary\, actigraphy\, and polysomnography. Challenges in this effort include the us e of mixtures of skewed distributions\, a large number of potential cluste ring variables\, and seeking clinically meaningful solutions. We present n ovel variable selection algorithms\, study them by simulation\, and demons trate our methods on the sleep data. This work is joint with Meredith Wall ace.\n LOCATION:https://researchseminars.org/talk/gbgstats/65/ END:VEVENT BEGIN:VEVENT SUMMARY:Linn Engström (KTH Royal Institute of Technology) DTSTART:20241023T111500Z DTEND:20241023T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/68 DESCRIPTION:Title: Computation of Robust Option Prices via Martingale Optimal Transport \nby Linn Engström (KTH Royal Institute of Technology) as part of Gothenb urg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nDuring the l ast decade there has been a rapid development of methods for computational ly addressing optimal transport problems\; motivated by applications withi n robust finance\, effort has also been made to generalize some of these t echniques to problems equipped with an additional martingale constraint. C omputationally solving multi-marginal martingale optimal transport problem s remains a challenging task though\, particularly for problems formulated with a large number of marginals.\n\nIn this talk I will give a brief int roduction to the martingale optimal transport problem and motivate why it is interesting from a mathematical finance point of view\, before presenti ng an efficient framework for solving a class of such multi-marginal probl ems computationally. The method combines the celebrated entropic regulariz ation with the exploitation of certain structures inherent in the problem\ , enabling fast computation of the optimal dual variables. I will also pro vide some examples that demonstrates the utility of our method in terms of computing model-independent bounds on the fair price of some exotic optio ns\, such as lookback options and Asian options. The talk is based on join t work with Sigrid Källblad and Johan Karlsson.\n LOCATION:https://researchseminars.org/talk/gbgstats/68/ END:VEVENT BEGIN:VEVENT SUMMARY:Marcel Wienöbst (University of Lübeck) DTSTART:20241106T090000Z DTEND:20241106T094500Z DTSTAMP:20260422T122728Z UID:gbgstats/70 DESCRIPTION:Title: Linear-Time Algorithms for Front-Door Adjustment in Causal Graphs\nb y Marcel Wienöbst (University of Lübeck) as part of Gothenburg statistic s seminar\n\nLecture held in MVL14.\n\nAbstract\nCausal effect estimation from observational data is a fundamental task in\nempirical sciences. It b ecomes particularly challenging when unobserved\nconfounders are involved in a system. Front-door adjustment constitutes a\nclassic method that allo ws identifying the causal effect even in the presence of\nlatent confoundi ng by using observed mediators. This talk presents a recent\nalgorithmic r esult in this area\, namely a linear-time algorithm for finding a\nfront-d oor adjustment set in a given causal graph. Its run-time is\nasymptoticall y optimal and improves on the previous state-of-the-art for this\ntask by a factor that grows cubically in the number of variables. Beyond this\nres ult\, the presentation explores fundamental algorithmic tools and techniqu es\nuseful for broader applications in causal inference.\n LOCATION:https://researchseminars.org/talk/gbgstats/70/ END:VEVENT BEGIN:VEVENT SUMMARY:Sebastian Persson (University of Gothenburg) DTSTART:20241127T121500Z DTEND:20241127T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/72 DESCRIPTION:Title: PEtab.jl - Efficient parameter estimation for dynamic models\nby Seb astian Persson (University of Gothenburg) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nOrdinary differential equa tions (ODEs) are commonly used to model dynamic processes such as biologic al networks. ODE models often contain unknown parameters that must be esti mated from data. From a statistical viewpoint\, this estimation is typical ly performed by computing a maximum likelihood estimate\, which boils down to solving a nonlinear optimization problem. In simple cases\, the likeli hood function can be easily coded using existing libraries in programming languages like Python and Julia. However\, for more complex scenarios—su ch as when the model includes events\, data is collected under various sim ulation conditions\, or the model should be at a steady state at time zero —correctly coding a likelihood function becomes time-consuming and error -prone. Moreover\, numerically fitting an ODE model to data can be computa tionally intensive\, potentially taking hours to days\, and the choice of ODE solver and gradient computation methods can drastically affect runtime . \n\n \nIn this talk\, I will discuss our software package PEtab.jl\, a J ulia package for setting up parameter estimation problems for dynamic mode ls. I will cover how PEtab.jl simplifies parameter estimation workflows an d present extensive benchmark results on how the choice of gradient method s and ODE solvers affects runtime. Lastly\, I will discuss how mechanistic models can be complemented with data-driven neural-network models to addr ess the shortcomings of each model type.\n LOCATION:https://researchseminars.org/talk/gbgstats/72/ END:VEVENT BEGIN:VEVENT SUMMARY:Lembris Njotto (University of Dar es Salaam) DTSTART:20241120T121500Z DTEND:20241120T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/73 DESCRIPTION:Title: Spatio-temporal modeling of malaria cases in Tanzania\nby Lembris Nj otto (University of Dar es Salaam) as part of Gothenburg statistics semina r\n\nLecture held in MVL14.\n\nAbstract\nMalaria continues to pose a signi ficant global health challenge\, affecting approximately 200 million indiv iduals annually and causing an estimated 600\,000 deaths worldwide. Enviro nmental factors are key drivers of malaria transmission dynamics\, influen cing disease patterns at local and regional scales. This talk focuses on d ata from Tanzania to explore the impact of climatic factors and vector con trol interventions on malaria incidence.\n\nUsing Standardized Incidence R atio (SIR) metrics and Bayesian spatio-temporal modeling approaches\, we a nalyze regionally aggregated monthly malaria cases\, stratified into two a ge groups: children under five and individuals aged five years and above. The models incorporate a Conditional Autoregressive (CAR) structure to cap ture spatial dependencies\, a second-order random walk (RW2) for temporal trends\, and independent and identically distributed (iid) random effects to account for unstructured spatial and temporal variability. Specific res ults on the influence of environmental factors\, including precipitation a nd temperature\, on malaria cases will be presented during the talk\, high lighting their role in transmission dynamics and informing targeted interv ention strategies.\n\nResults are not yet published\, please make them con fidential.\n LOCATION:https://researchseminars.org/talk/gbgstats/73/ END:VEVENT BEGIN:VEVENT SUMMARY:Martin Voigt Vejling (Aalborg University) DTSTART:20241121T100000Z DTEND:20241121T110000Z DTSTAMP:20260422T122728Z UID:gbgstats/74 DESCRIPTION:Title: Conformal multiple Monte Carlo testing with a view to spatial statistics \nby Martin Voigt Vejling (Aalborg University) as part of Gothenburg s tatistics seminar\n\nLecture held in MVL14.\n\nAbstract\nMonte Carlo tests are popular for their convenience\, as they allow the computation of vali d p-values even when test statistics with known and tractable distribution s are unavailable. When performing multiple Monte Carlo tests\, it is esse ntial to adjust the testing procedure to maintain control of the type I er ror\, and some of such techniques pose requirements on the joint distribut ion of the p-values\, for instance independence. A straightforward approac h to get independent p-values\, is to compute the p-values for each hypoth esis in parallel\, however\, this imposes a substantial computational burd en. We highlight in this work that the problem of testing multiple data sa mples against the same null hypothesis is an instance of conformal outlier detection. Leveraging this insight enables a more efficient multiple Mont e Carlo testing procedure\, avoiding excessive simulations while still ens uring exact control over the false discovery rate. Through numerical exper iments on point patterns we investigate the performance of this proposed c onformal multiple Monte Carlo testing (CMMCTest) procedure.\n LOCATION:https://researchseminars.org/talk/gbgstats/74/ END:VEVENT BEGIN:VEVENT SUMMARY:Aila Särkkä (Chalmers University of Technology & University of G othenburg) DTSTART:20241211T121500Z DTEND:20241211T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/75 DESCRIPTION:Title: Analysis of point patterns observed with errors: some examples\nby A ila Särkkä (Chalmers University of Technology & University of Gothenburg ) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAb stract\nMany natural systems are observed as point patterns in time\, spac e\, or space and time. Examples include plant and cellular systems\, anima l colonies\, wildfires\, and galaxies. In practice\, the locations of the points are not always observed correctly. However\, in the point process l iterature\, little attention has been paid to the issue of errors in the l ocations of points. In this talk\, we discuss how the observed point patte rn may deviate from the actual point pattern\, review methods and models t hat exist to handle such deviations\, and give some examples of data obser ved with errors. \n\nBased on joint work with Peter Guttorp (Norwegian Com puting Center)\, Janine Illian (University of Glasgow)\, Joel Kostensalo ( Natural Resources Institute Finland (Luke)\, Mikko Kuronen (Luke)\, Mari M yllymäki (Luke)\, and Thordis Thorarinsdottir (University of Oslo).\n LOCATION:https://researchseminars.org/talk/gbgstats/75/ END:VEVENT BEGIN:VEVENT SUMMARY:Henrik Häggström (Chalmers University of Technology & University of Gothenburg) DTSTART:20250122T121500Z DTEND:20250122T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/76 DESCRIPTION:Title: Simulation-based inference for stochastic nonlinear mixed-effects models with applications in systems biology\nby Henrik Häggström (Chalmers University of Technology & University of Gothenburg) as part of Gothenbur g statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nWe propose a n ovel methodology for Bayesian inference in hierarchical mixed-effects mode ls. We construct a framework that is highly scalable\, where amortized app roximations to the likelihood and the parameters posterior are first obtai ned\, and these are rapidly refined for each individual dataset\, to ultim ately approximate the parameters posterior across many individuals. The fr amework introduced is easily trainable\, as it uses mixture of experts but without neural networks\, leading to parsimonious yet expressive surrogat e models of the likelihood and the posterior. The methodology is exemplifi ed via stochastic differential equation mixed-effects models\, that are hi ghly relevant in systems biology\, but the methodology is general and can accommodate other types of stochastic and deterministic models. We compare our approximate inference with exact pseudomarginal inference and show th at our methodology is fast and competitive.\n LOCATION:https://researchseminars.org/talk/gbgstats/76/ END:VEVENT BEGIN:VEVENT SUMMARY:Julia Jansson (Chalmers University of Technology & University of G othenburg) DTSTART:20250109T083000Z DTEND:20250109T103000Z DTSTAMP:20260422T122728Z UID:gbgstats/77 DESCRIPTION:Title: Licentiate seminar: Statistical Properties of Point Process Learning for Gibbs Processes\nby Julia Jansson (Chalmers University of Technology & University of Gothenburg) as part of Gothenburg statistics seminar\n\nLe cture held in Pascal\, Hörsalsvägen 1.\n\nAbstract\nThis thesis studies Point Process Learning (PPL)\, which is a novel statistical learning frame work that uses point process cross-validation and point process prediction errors\, and includes different hyperparameters. Specifically\, statistic al properties of PPL are explored\, in the context of Gibbs point processe s. Paper 1 demonstrates PPL’s advantages over pseudolikelihood\, which i s a state-of-the-art parameter estimation method and a special case of Tak acs- Fiksel estimation (TF)\, with particular focus on Gibbs hard-core pro cesses. Paper 2 compares PPL to TF\, and shows that TF is a special case o f PPL\, when the cross-validation scheme tends to leave-one-out cross-vali dation. In addition\, Paper 2 shows that for four common Gibbs models\, na mely Poisson\, hard-core\, Strauss and Geyer saturation processes\, one ca n choose hyperparameters so that PPL outperforms TF in terms of mean squar e error.\n\nIn Paper 1 and 2\, parameter estimation with PPL is done by mi nimizing loss functions\, while Paper 3 explores an alternative approach t o PPL\, namely estimating equations. Further\, statistical properties of t he parameter estimator are derived in Paper 3\, such as consistency and as ymptotic normality for large samples\, as well as bias and variance for sm all samples. It is concluded that the estimating equation approach is not feasible for PPL\, whereby the original loss function-based approach is pr eferred. Moving on\, Paper 3 then provides a theoretical foundation for th e loss functions through an empirical risk formulation.\n\nTo conclude\, P PL is shown to be a flexible and robust competitor to state-of-the-art met hods for parameter estimation.\n\nRoom: Pascal\, Hörsalsvägen 1\n LOCATION:https://researchseminars.org/talk/gbgstats/77/ END:VEVENT BEGIN:VEVENT SUMMARY:Alexey Lindo (University of Glasgow) DTSTART:20250115T121500Z DTEND:20250115T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/78 DESCRIPTION:Title: Probability-Generating Function Kernels for Spherical Data\nby Alexe y Lindo (University of Glasgow) as part of Gothenburg statistics seminar\n \nLecture held in MVL14.\n\nAbstract\nIn this talk\, I will introduce the class of probability-generating function (PGF) kernels\, a novel approach to spherical data analysis. PGF kernels generalize radial basis function ( RBF) kernels and are supported on the unit hypersphere\, making them well- suited for tasks involving spherical data. I will discuss their unique pro perties\, demonstrate a semi-parametric learning algorithm for fitting the se kernels\, and showcase their application in Gaussian processes and deep kernel learning. Through examples and comparisons\, I will highlight the advantages of PGF kernels over existing methods.\n LOCATION:https://researchseminars.org/talk/gbgstats/78/ END:VEVENT BEGIN:VEVENT SUMMARY:Sara Hamis (Uppsala University) DTSTART:20250423T111500Z DTEND:20250423T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/79 DESCRIPTION:Title: Predicting and controlling cell systems that generate spatio-temporal po int patterns\nby Sara Hamis (Uppsala University) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nRecent technolo gical advances have resulted in a multitude of spatio-temporal cell imagin g data. These can be translated into spatio-temporal point patterns in whi ch points represent cells. Such data hold rich information about how cells act and interact\, much of which is not extractable through data analysis alone. Therefore\, to identify\, predict and control cell systems that ge nerate spatio-temporal patterns\, we propose using two unified classes of mathematical models: spatio-temporal point processes (STPPs) and spatial c umulant models (SCMs). SCMs are population models formulated by differenti al equations that approximate the dynamics of two STPP-generated summary s tatistics: first-order spatial cumulants (densities)\, and second-order sp atial cumulants (spatial covariances). In this talk\, I’ll demonstrate t hat (1) SCMs can capture STPP-generated density dynamics\, even when mean- field population models (MFPMs) fail to do so\, and (2) SCM-informed treat ment strategies outperform MFPM-informed strategies in terms of inhibiting population growths. Overall\, our work demonstrates that SCMs provide a p romising framework in which to study ecological systems that generate spat io-temporal point patterns in cell biology and beyond.\n LOCATION:https://researchseminars.org/talk/gbgstats/79/ END:VEVENT BEGIN:VEVENT SUMMARY:Alvaro Köhn-Luque (University of Oslo) DTSTART:20250430T090000Z DTEND:20250430T094500Z DTSTAMP:20260422T122728Z UID:gbgstats/80 DESCRIPTION:Title: Phenotypic deconvolution of cancer cell populations\nby Alvaro Köhn -Luque (University of Oslo) as part of Gothenburg statistics seminar\n\nLe cture held in MVL15.\n\nAbstract\nTumor heterogeneity is an important driv er of treatment failure in cancer\, as therapies often select for drug-tol erant or drug-resistant cellular subpopulations that drive tumor growth an d recurrence. Profiling the drug-response heterogeneity of tumor samples u sing traditional genomic deconvolution methods has yielded limited results \, due in part to the imperfect mapping between genomic variation and func tional characteristics. In this seminar\, I will demonstrate how to levera ge mechanistic population modeling to develop a statistical framework for profiling phenotypic heterogeneity from standard drug-screen data on bulk tumor samples. This approach allows us to reliably identify tumor subpopul ations exhibiting differential drug responses and estimate their drug sens itivities and frequencies within the bulk population. I will discuss the a dvantages and disadvantages of using deterministic versus stochastic birth -death population models. These methods are applied to synthetically gener ated cell populations\, mixed cell-line in vitro experiments\, and multipl e myeloma patient samples.\n LOCATION:https://researchseminars.org/talk/gbgstats/80/ END:VEVENT BEGIN:VEVENT SUMMARY:Calum Gabbutt (Imperial College London) DTSTART:20250507T111500Z DTEND:20250507T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/81 DESCRIPTION:Title: Timing copy number alterations in Barrett’s Oesophagus using hierarchi cal Bayesian models\nby Calum Gabbutt (Imperial College London) as par t of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\n The accumulation of somatic copy number alterations (CNAs) is a key genomi c risk factor in the progression from Barrett’s Oesophagus (a pre-malign ant condition\, BO) to oesophageal adenocarcinoma (OAC). However\, the tim ing and evolutionary dynamics of these CNAs have remained elusive. In this talk\, I will introduce CARBINE (Copy number AlteRation timing with Bayes ian Inference and Neutral Evolution)\, a hierarchical Bayesian framework d esigned to infer the calendar-time occurrence of CNAs from deep whole-geno me sequencing data. CARBINE leverages molecular clock signals and clonal e volutionary theory to estimate patient-specific mutation rates\, clonal gr owth dynamics\, and the timing of genomic events from single-timepoint sam ples. Using this method\, we find that critical alterations\, such as whol e-genome doubling and TP53 inactivation\, often occur decades before cance r diagnosis\, including during early life\, and are followed by long perio ds of indolent clonal expansion. Furthermore\, we show that the rate of CN A accumulation—estimated from single snapshots—outperforms overall bur den as a predictor of progression to OAC. This new insight into the tempor al evolution of BO underscores the potential of early-life genomic profili ng to stratify cancer risk and informs strategies for early intervention.\ n LOCATION:https://researchseminars.org/talk/gbgstats/81/ END:VEVENT BEGIN:VEVENT SUMMARY:Rasmus Waagepetersen (Aalborg University) DTSTART:20250110T100000Z DTEND:20250110T104500Z DTSTAMP:20260422T122728Z UID:gbgstats/82 DESCRIPTION:Title: Point process approaches to study clustering of industry locations\n by Rasmus Waagepetersen (Aalborg University) as part of Gothenburg statist ics seminar\n\nLecture held in MVL15.\n\nAbstract\nIn this talk we will di scuss various point process approaches to study clustering of industry loc ations. Industries (shops\, firms\, supermarkets\, factories...) can be of various types and clustering (or possibly repulsion) could happen within industries of the same type or between industries of different types. We r eview a seminal contribution in spatial econometrics and discuss its relat ion to recent semi-parametric point process models including semi-parametr ic log Gaussian Cox processes and semi-parametric Markov point processes. For the semi-parametric models we in particular consider how parameter est imates can be obtained using certain conditional composite likelihoods.\n\ nRoom: MVL15\n LOCATION:https://researchseminars.org/talk/gbgstats/82/ END:VEVENT BEGIN:VEVENT SUMMARY:Elizabeth Baker (DTU) DTSTART:20250611T111500Z DTEND:20250611T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/83 DESCRIPTION:Title: Conditioning diffusion processes with score matching methods\nby Eli zabeth Baker (DTU) as part of Gothenburg statistics seminar\n\nLecture hel d in MVL14.\n\nAbstract\nIn stochastic optimal control and conditional gen erative modelling\, a central computational task is to modify a reference diffusion process to maximise a given terminal-time reward. Most existing methods require this reward to be differentiable\, using gradients to stee r the diffusion towards favourable outcomes. However\, in many practical s ettings\, like diffusion bridges\, the reward is singular\, taking an infi nite value if the target is hit and zero otherwise. We introduce a novel f ramework\, based on Malliavin calculus and path-space integration by parts \, that enables the development of methods robust to such singularities. T his allows our approach to handle a broad range of applications\, includin g classification\, diffusion bridges\, and conditioning without the need f or artificial observational noise.\n LOCATION:https://researchseminars.org/talk/gbgstats/83/ END:VEVENT BEGIN:VEVENT SUMMARY:Mattias Byléhn (Chalmers University of Technology & University of Gothenburg) DTSTART:20250528T111500Z DTEND:20250528T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/84 DESCRIPTION:Title: PhD defence: Hyperuniformity and hyperfluctuations for random measures o n Euclidean and non-Euclidean spaces\nby Mattias Byléhn (Chalmers Uni versity of Technology & University of Gothenburg) as part of Gothenburg st atistics seminar\n\nLecture held in Euler.\n\nAbstract\nReserved slot beca use of https://www.chalmers.se/en/current/calendar/mv-doctoral-thesis-matt ias-byhlen/\n LOCATION:https://researchseminars.org/talk/gbgstats/84/ END:VEVENT BEGIN:VEVENT SUMMARY:Vincent Molin (Chalmers University of Technology & University of G othenburg) DTSTART:20250416T111500Z DTEND:20250416T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/85 DESCRIPTION:Title: Controlled stochastic processes for simulated annealing\nby Vincent Molin (Chalmers University of Technology & University of Gothenburg) as pa rt of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\ nSimulated annealing solves optimization problems by means of a random wal k in an energy landscape based on the objective function and a temperature parameter. By slowly decreasing the temperature\, the algorithm converges to the global optimal solution\, also for nonconvex functions. However\, if the temperature is decreased too quickly\, this procedure often gets st uck in local minima. To overcome this\, we here present a new perspective on simulated annealing. More precisely\, we consider the cooling landscape as a curve of probability measures and prove that there exists a minimal norm velocity field which solves the continuity equation. The latter is a differential equation which governs the evolution of the aforementioned cu rve. The solution is the weak gradient of an integrable function\, which i s in line with the interpretation of the velocity field as a derivative of optimal transport maps. We also show that controlling stochastic annealin g processes by superimposing this velocity field would allow them to follo w arbitrarily fast cooling schedules. Based on these findings\, we design novel interacting particle based optimization methods\, convergent optimal transport based approximations to this control\, that accelerate simulate d annealing processes. This acceleration behavior is also validated on a n umber of numerical experiments.\n LOCATION:https://researchseminars.org/talk/gbgstats/85/ END:VEVENT BEGIN:VEVENT SUMMARY:Adélie Erard (Université Paris Cité) DTSTART:20250319T121500Z DTEND:20250319T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/86 DESCRIPTION:Title: A method for estimating population growth at a local scale. An applicati on using French breeding bird surveys data.\nby Adélie Erard (Univers ité Paris Cité) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nEstimating population growth at a local scale is c rucial for understanding ecological dynamics and informing conservation ef forts. In this work\, we propose a novel methodology that uses marked poin t processes to model bird populations as a spatial process of unknown inte nsity influenced by environmental factors. The population is represented a s a point process $\\mathcal{P}$\, while observations are obtained through a thinning mechanism using a homogeneous Poisson birth-and-death process $\\mathcal{O}_t$. This approach accounts for both spatial dependence\, suc h as Cox or Gibbs processes\, and temporal variations by focusing on inter sections of observed areas over consecutive time points.\n\n\nBy applying stabilization theory\, we demonstrate convergence properties and ensure ro bust local estimations of population variation. The ability to estimate po pulation dynamics at fine spatial scales distinguishes this approach\, mak ing it particularly suited to ecological applications.\n\nAn implementatio n using French breeding bird survey data highlights its potential to captu re localized trends and advance biodiversity monitoring.\n LOCATION:https://researchseminars.org/talk/gbgstats/86/ END:VEVENT BEGIN:VEVENT SUMMARY:Nils Grimbeck (Chalmers University of Technology & University of G othenburg) DTSTART:20250527T083000Z DTEND:20250527T091500Z DTSTAMP:20260422T122728Z UID:gbgstats/87 DESCRIPTION:Title: What do we see when we look at art? - Modelling of eye-movements in art perception\nby Nils Grimbeck (Chalmers University of Technology & Univ ersity of Gothenburg) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nEye movements during art perception have been extensively studied over the past century\, as they provide insights into perceptual\, evaluative\, and cognitive processes. Although several theore tical frameworks have been proposed\, it is only recently that spatial sta tistics have begun to explore gaze patterns\, specifically by modelling fi xation locations as spatial point patterns arising from spatio-temporal po int processes. Inspired by the simple model of eye-movements proposed by Y litalo et al. in 2016\, we propose a stochastic model that incorporates bo th the semi-conscious transitions between regions of interest (ROIs) in a painting\, as well as the semi-random eye movements that occur while regis tering visual information within these regions during the first 30 seconds of art perception. \n \n\nUsing eye-tracking data from twenty subjects on six paintings\, we apply mean-shift clustering to identify ROIs in eac h painting based on the intensity of fixation points. A Markov chain is su bsequently used to model the transitions between these regions and based o n the model proposed by Ylitalo et al. we use the estimated intensity and saccade length distribution to model the placement of fixations within eac h ROI. Using this modelling approach\, we analyse the dynamics of eye move ments during the initial 30 seconds of art perception and to assess the ro bustness of our modelling assumptions across six diverse paintings and art istic styles.\n LOCATION:https://researchseminars.org/talk/gbgstats/87/ END:VEVENT BEGIN:VEVENT SUMMARY:Marco Tarantino (University of Palermo) DTSTART:20250326T121500Z DTEND:20250326T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/88 DESCRIPTION:Title: Using a Neural Network approach and Starspots dependent models to predic t Effective Temperatures and Ages of young stars\nby Marco Tarantino ( University of Palermo) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nThis study presents a statistical approach to accurately predict the effective temperatures of pre-main sequence stars\ , which are necessary for determining stellar ages using the isochrone met hodology and cutting-age starspots-dependent models. By training a Neural Network model on high-quality spectroscopic temperatures from the Gaia-ESO Survey as the response variable\, and using photometric data from Gaia DR 3 and 2MASS catalogs as explanatory variables\, we implemented a methodolo gy to accurately derive the effective temperatures of much larger populati ons of stars for which only photometric data are available. The model demo nstrated robust performance for low-mass stars with temperatures below 700 0 K\, including young stars\, the primary focus of this work. Predicted t emperatures were employed to construct Hertzsprung-Russell diagrams and to predict stellar ages of different young clusters and star forming regions through isochrone interpolation\, achieving excellent agreement with spec troscopic-based ages and literature values derived from model-independent methods like lithium equivalent widths. The inclusion of starspot evolutio nary models improved the age predictions\, providing a more accurate descr iption of stellar properties. Additionally\, the results regarding the eff ective temperature and age predictions of the young clusters provide evide nces of the presence of intrinsic age spreads in the youngest clusters\, s uggesting multiple formation events over time.\n LOCATION:https://researchseminars.org/talk/gbgstats/88/ END:VEVENT BEGIN:VEVENT SUMMARY:Jeffrey Steif (Chalmers University of Technology & University of G othenburg) DTSTART:20250903T111500Z DTEND:20250903T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/89 DESCRIPTION:Title: Is your coin rational?\nby Jeffrey Steif (Chalmers University of Tec hnology & University of Gothenburg) as part of Gothenburg statistics semin ar\n\nLecture held in MVL14.\n\nAbstract\nOne tosses a coin with an unknow n parameter p a large number of times and then you have to guess whether p is rational or irrational. Can you do it? The answer is related to variou s things such as\nthe so-called Baire Category Theorem as well as well-app roximability of irrationals by rationals \nin elementary number theory.\n LOCATION:https://researchseminars.org/talk/gbgstats/89/ END:VEVENT BEGIN:VEVENT SUMMARY:Ruben Seyer (Chalmers University of Technology & University of Got henburg) DTSTART:20250521T111500Z DTEND:20250521T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/90 DESCRIPTION:Title: Creating non-reversible rejection-free samplers by rebalancing skew-bala nced Markov jump processes\nby Ruben Seyer (Chalmers University of Tec hnology & University of Gothenburg) as part of Gothenburg statistics semin ar\n\nLecture held in MVL14.\n\nAbstract\nMarkov chain sampling methods fo rm the backbone of modern computational statistics. However\, many popular methods are prone to random walk behavior\, i.e.\, diffusion-like explora tion of the sample space\, leading to slow mixing that requires intricate tuning to alleviate. Non-reversible samplers can resolve some of these iss ues. We introduce a device that turns jump processes that satisfy a skew-d etailed balance condition for a reference measure into a process that samp les a target measure that is absolutely continuous with respect to the ref erence measure. The resulting sampler is rejection-free\, non-reversible\, and continuous-time. As an example\, we apply the device to Hamiltonian d ynamics discretized by the leapfrog integrator\, resulting in a rejection- free non-reversible continuous-time version of Hamiltonian Monte Carlo (HM C). We prove the geometric ergodicity of the resulting sampler under certa in convexity conditions\, and demonstrate its qualitatively different beha vior to HMC through numerical examples.\n LOCATION:https://researchseminars.org/talk/gbgstats/90/ END:VEVENT BEGIN:VEVENT SUMMARY:Philip Gerlee (Chalmers University of Technology & University of G othenburg) DTSTART:20250618T111500Z DTEND:20250618T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/91 DESCRIPTION:Title: Evaluation of respiratory disease hospitalisation forecasts using synthe tic outbreak data\nby Philip Gerlee (Chalmers University of Technology & University of Gothenburg) as part of Gothenburg statistics seminar\n\nL ecture held in MVL14.\n\nAbstract\nForecasts of hospitalisations of infect ious diseases play an important role for allocating healthcare resources d uring epidemics and pandemics. Large-scale analysis of model forecasts dur ing the COVID-19 pandemic has shown that the model rank distribution with respect to accuracy is heterogeneous and that ensemble forecasts have the highest average accuracy. Building on that work we generated a maximally d iverse synthetic dataset of 324 different hospitalisation time-series that correspond to different disease characteristics and public health respons es. We evaluated forecasts from 14 component models and 6 different ensemb les. Our results show that component model accuracy was heterogeneous and varied depending on the current rate of disease transmission. Going from 7 day to 14 day forecasts mechanistic models improved in relative accuracy compared to statistical models. A novel adaptive ensemble method outperfor ms all other ensembles\, but is closely followed by a median ensemble. We also investigated the relationship between ensemble error and variability of component forecasts and show that the coefficient of variation is predi ctive of future error. Lastly\, we validated the results on data from the COVID-19 pandemic in Sweden. Our findings have the potential to improve ep idemic forecasting\, in particular the ability to assign confidence to ens emble forecasts at the time of prediction based on component forecast vari ability.\n LOCATION:https://researchseminars.org/talk/gbgstats/91/ END:VEVENT BEGIN:VEVENT SUMMARY:Fima Klebaner (Monash University) DTSTART:20250710T111500Z DTEND:20250710T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/92 DESCRIPTION:Title: Emergence of populations. H(W) theory.\nby Fima Klebaner (Monash Uni versity) as part of Gothenburg statistics seminar\n\nLecture held in MVL15 .\n\nAbstract\nWe study how populations emerge when starting with just a f ew individuals\, maybe only one\, and then growing to its (large) carrying capacity K. We prove an old conjecture and suggest new approximations.\n\ nThe talk is based on a number of papers with: Andrew Barbour\, Pavel Chig ansky\, Peter Jagers\, Kais Hamza\, and PhD students Jeremy Baker and Naor Bauman.\n LOCATION:https://researchseminars.org/talk/gbgstats/92/ END:VEVENT BEGIN:VEVENT SUMMARY:Oskar Allerbo (KTH Royal Institute of Technology) DTSTART:20251008T111500Z DTEND:20251008T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/93 DESCRIPTION:Title: Is supervised learning really that different from unsupervised?\nby Oskar Allerbo (KTH Royal Institute of Technology) as part of Gothenburg st atistics seminar\n\nLecture held in MVL14.\n\nAbstract\nWe demonstrate how supervised learning can be decomposed into a two-stage procedure\, where (1) all model parameters are selected in an unsupervised manner\, and (2) the outputs y are added to the model\, without changing the parameter valu es. This is achieved by a new model selection criterion that - in contrast to cross-validation - can be used also without access to y. For linear ri dge regression\, we bound the asymptotic out-of-sample risk of our method in terms of the optimal asymptotic risk. We also demonstrate on real and s ynthetic data that versions of linear and kernel ridge regression\, smooth ing splines\, and neural networks\, which are trained without access to y\ , perform similarly to their standard y-based counterparts. Hence\, our re sults suggest that the difference between supervised and unsupervised lear ning is less fundamental than it may appear.\nJoint work with Thomas B. Sc hön.\n LOCATION:https://researchseminars.org/talk/gbgstats/93/ END:VEVENT BEGIN:VEVENT SUMMARY:Filip Tronarp (Lund University) DTSTART:20251119T121500Z DTEND:20251119T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/94 DESCRIPTION:Title: A Recursive Theory of Variational State Estimation: The Dynamic Programm ing Approach\nby Filip Tronarp (Lund University) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nIn this talk\, we discuss the variational inference problem in partially observed Markov processes from the dynamic programming perspective. \nThis leads to a back ward and a forward recursion for certain value functionals\, which are clo sely connected to the corresponding recursions from classical Bayesian sta te estimation theory. Namely\, the backward value functional is a lower bo und on the "backward filter" and the forward value functional is a lower b ound on the unnormalized filtering density. The two recursions can also be combined yielding a variational two-filter formula.\nWhat results is a va riational state estimation theory that is completely analogous to the clas sical Bayesian state estimation theory. \nThe theory is applied to a jump Gauss-Markov regression problem\, where closed form solutions to the value functional recursions can be obtained.\n LOCATION:https://researchseminars.org/talk/gbgstats/94/ END:VEVENT BEGIN:VEVENT SUMMARY:Johannes Borgqvist (Chalmers) DTSTART:20251015T111500Z DTEND:20251015T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/95 DESCRIPTION:Title: HeMiTo-dynamics: a characterization of mammalian prion toxicity using no n-dimensionalization\, linear stability and perturbation analyses\nby Johannes Borgqvist (Chalmers) as part of Gothenburg statistics seminar\n\n Lecture held in MVL14.\n\nAbstract\nPrion-like proteins play crucial parts in biological processes in organisms ranging from yeast to humans. For in stance\, many neurodegenerative diseases are believed to be caused by the production of prion-like proteins in neural tissue. As such\, understandin g the dynamics of prion-like protein production is a vital step toward tre ating neurodegenerative disease. Mathematical models of prion-like protein dynamics show great promise as a tool for predicting disease trajectories and devising better treatment strategies for prion-related diseases. Here in\, we investigate a generic model for prion-like dynamics consisting of a class of non-linear ordinary differential equations (ODEs)\, establishin g constraints through a linear stability analysis that enforce the expecte d properties of mammalian prion-like toxicity. Furthermore\, we identify t hat prion toxicity evolves through three distinct phases for which we prov ide analytical descriptions using perturbation analyses. Specifically\, pr ion-toxicity is initially characterized by the healthy phase\, where the d ynamics are dominated by the healthy form of prions\, thereafter the syste m enters the mixed phase\, where both healthy and toxic prions interact\, and lastly\, the system enters the toxic phase\, where toxic prions domina te\, and we refer to these phases as HeMiTo-dynamics. These findings hold the potential to aid researchers in developing precise mathematical models for prion-like dynamics\, enabling them to better understand underlying m echanisms and devise effective treatments for prion-related diseases.\n\nA t this point in time\, the work has been solely focused on analysing a cla ss of mathematical models of prion diseases. Next\, the plan is to start t wo new projects involving experimental data from medical collaborators. In short\, these projects involve identifying an unknown conversion function in our class of prion models using time series data in combination with p hysics informed neural networks\, as well as spatial modelling of how prio ns are distributed over time in diseased brains. The main aim of this talk is to start a discussion about these collaboration projects\, and any inp ut would be greatly appreciated.\n\nThe slide-based presentation is made i n beamer\, and I will bring my own laptop to the presentation.\n LOCATION:https://researchseminars.org/talk/gbgstats/95/ END:VEVENT BEGIN:VEVENT SUMMARY:Kasper Bågmark (Chalmers) DTSTART:20260211T121500Z DTEND:20260211T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/96 DESCRIPTION:Title: High-dimensional Bayesian filtering through deep density approximation\nby Kasper Bågmark (Chalmers) as part of Gothenburg statistics seminar \n\nLecture held in MVL14.\n\nAbstract\nIn this work\, we benchmark two re cently developed deep density methods for nonlinear filtering. Starting fr om the Fokker--Planck equation with Bayes updates\, we model the filtering density of a discretely observed SDE. The two filters: the deep splitting filter and the deep BSDE filter\, are both based on Feynman--Kac formulas \, Euler--Maruyama discretizations and neural networks. The two methods ar e extended to logarithmic formulations providing sound and robust implemen tations in increasing state dimension. Comparing to the classical particle filters and ensemble Kalman filters\, we benchmark the methods on numerou s examples. In the low-dimensional examples the particle filters work well \, but when we scale up to a partially observed $100$-dimensional Lorenz-9 6 model the particle-based methods fail and the logarithmic deep density m ethod prevails. In terms of computational efficiency\, the deep density me thods reduce inference time by roughly two to five orders of magnitude rel ative to the particle-based filters.\n LOCATION:https://researchseminars.org/talk/gbgstats/96/ END:VEVENT BEGIN:VEVENT SUMMARY:Sandra Barman (RISE Research Institutes of Sweden) DTSTART:20260204T121500Z DTEND:20260204T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/97 DESCRIPTION:Title: Correlating structure and material properties in soft materials\nby Sandra Barman (RISE Research Institutes of Sweden) as part of Gothenburg s tatistics seminar\n\nLecture held in MVL14.\n\nAbstract\nThis talk focuses on how statistics and machine learning can be used to correlate the nano- and microstructure of soft materials with their material properties. We w ork with different application areas where this is relevant\, including pa ckaging and barrier materials\, hygiene products\, pharmaceuticals\, and f ood.\n\nTo develop models that map the relationship between a material’s structure and its functional properties\, we combine:\n\n
    \n
  1. meth ods for material imaging to understand what the structure looks like\, ran ging from indirect methods such as X-ray scattering to direct imaging in 2 D and 3D\, with or without a time component\,
    2. \n
  2. models for repli cating and exploring 3D material structure using spatial statistics and ge nerative AI\,
    3. \n
  3. numerical simulation of functional properties su ch as fluid and gas transport\, and
    4. \n
  4. statistical and machine le arning models that connect structure to functional properties.
    5. \n
\n\nI will in this talk give an overview of some ongoing projects which ar e done in collaboration between RISE\, the Department of Mathematical Scie nces and Department of Physics at Chalmers\, Chalmers Industriteknik\, and industrial partners such as Tetra Pak\, AstraZeneca\, and Essity.\n LOCATION:https://researchseminars.org/talk/gbgstats/97/ END:VEVENT BEGIN:VEVENT SUMMARY:Taisiia Morozova (Uppsala University) DTSTART:20251105T121500Z DTEND:20251105T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/98 DESCRIPTION:Title: Multi-Agent Reinforcement Learning for Buffered Cellular Networks\nb y Taisiia Morozova (Uppsala University) as part of Gothenburg statistics s eminar\n\nLecture held in MVL14.\n\nAbstract\nWe study the use of multi-ag ent reinforcement learning (MARL) for buffered cellular networks\, where b ase stations are modelled as independent agents making transmission decisi ons under interference and delay constraints. The network is described thr ough a stochastic geometry framework with Poisson-distributed base station s and users\, and buffers capturing traffic arrivals and service dynamics. To handle the interactions between agents\, we employ a mean-field approx imation\, so that each agent responds to an aggregate distribution of its neighbours’ states and actions. The learning problem is formulated via m ean-field Q-learning\, where the objective is to improve network capacity while controlling delays. Initial experiments show convergence of the Q-fu nctions for several agents\, suggesting that the approach is well-suited t o this setting.\n LOCATION:https://researchseminars.org/talk/gbgstats/98/ END:VEVENT BEGIN:VEVENT SUMMARY:Mathis Rost (Chalmers) DTSTART:20251022T111500Z DTEND:20251022T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/99 DESCRIPTION:Title: Void Probabilities and Likelihood Approximation for Gibbs Processes\ nby Mathis Rost (Chalmers) as part of Gothenburg statistics seminar\n\nLec ture held in MVL14.\n\nAbstract\nWhen fitting a model to data\, one would ideally like to use maximum likelihood estimation\, due to its nice statis tical properties. Unfortunately\, the likelihood function\nof a general Gi bbs point process is typically not tractable\, due to the associated norma lizing constant. This has led to the development of a range of alternative methods\,\nsuch as Takacs-Fiksel estimation (including its special case p seudolikelihood estimation) and Point Process Learning.\nLeveraging recent probabilistic results for Gibbs processes\, in this talk we present an\na pproach to perform approximate likelihood estimation for Gibbs processes. Specifically\, we show that the likelihood function can be expressed compl etely in terms of\nthe Papangelou conditional intensity\, which is typical ly known and tractable. This\nnew likelihood representation involves an in finite series expansion\, and we discuss\ndifferent ways of approximating it\, and thereby the likelihood function. We further\ndiscuss how this pla ys out in certain models and compare it to the state-of-the-art.\n LOCATION:https://researchseminars.org/talk/gbgstats/99/ END:VEVENT BEGIN:VEVENT SUMMARY:Simon Olsson (Chalmers) DTSTART:20260121T121500Z DTEND:20260121T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/100 DESCRIPTION:Title: Transferable Implicit Transfer Operators\nby Simon Olsson (Chalmers ) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAb stract\nIn this talk\, I will outline our approach to use deep generative models to learn weak solutions to the Langevin equations with long time ho rizons. E.g. given an initial condition\, $x_0$\, learn the transition den sity $p_t(x_t\\mid x_0)$\, where $t$ is orders of magnitude larger than th e usual numerical integration step. The context of this work is the $\\tex tit{sampling problem}$ from molecular dynamics\, an important method in ch emistry\, physics\, and biology\, that faces slow mixing. I will give nume rous empirical examples of the successful application of this approach in molecular dynamics.\n LOCATION:https://researchseminars.org/talk/gbgstats/100/ END:VEVENT BEGIN:VEVENT SUMMARY:Fabio Frommer (University of Mainz) DTSTART:20251210T121500Z DTEND:20251210T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/101 DESCRIPTION:Title: The inverse Henderson problem from statistical mechanics for multi-spec ies models\nby Fabio Frommer (University of Mainz) as part of Gothenbu rg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nThe inverse H enderson problem of statistical mechanics is the theoretical foundation fo r many bottom-up coarse-graining techniques for the numerical simulation o f complex soft matter physics. This inverse problem concerns classical par ticles in continuous space interacting according to a pair potential depen ding on the distance of the particles. Roughly stated\, it asks for the in teraction potential given the pair correlation function of the system. We show that the solution to this inverse problem is unique and can be rewrit ten as a minimization problem for a certain relative entropy functional. Lastly\, we show how this framework can be adapted to multi-species model s using marked Gibbs measures.\n LOCATION:https://researchseminars.org/talk/gbgstats/101/ END:VEVENT BEGIN:VEVENT SUMMARY:Zheng Zhao (Linköping University) DTSTART:20260218T121500Z DTEND:20260218T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/102 DESCRIPTION:Title: Diffusion differentiable resampling\nby Zheng Zhao (Linköping Univ ersity) as part of Gothenburg statistics seminar\n\nLecture held in MVL14. \n\nAbstract\nThis work is concerned with differentiable resampling in the context of sequential Monte Carlo (e.g.\, particle filtering). We propose a new informative resampling method that is instantly pathwise differenti able\, based on an ensemble score diffusion model. We prove that our diffu sion resampling method provides a consistent estimate to the resampling di stribution\, and we show by experiments that it outperforms the state-of-t he-art differentiable resampling methods when used for stochastic filterin g and parameter estimation. Implementations are available online at https: //github.com/zgbkdlm/diffres\n LOCATION:https://researchseminars.org/talk/gbgstats/102/ END:VEVENT BEGIN:VEVENT SUMMARY:Viktor Nilsson (KTH) DTSTART:20251028T100000Z DTEND:20251028T104500Z DTSTAMP:20260422T122728Z UID:gbgstats/103 DESCRIPTION:Title: New results in the large deviations of Schrödinger bridges\nby Vik tor Nilsson (KTH) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nIn a recent paper\, we show a large deviation prin ciple for certain sequences of static Schrödinger bridges\, typically mot ivated by a scale-parameter decreasing towards zero\, extending existing l arge deviation results to cover a wider range of reference processes. Our results provide a theoretical foundation for studying convergence of such Schrödinger bridges to their limiting optimal transport plans. Recently\, Bernton et al. established a large deviation principle\, in the small-noi se limit\, for fixed-cost entropic optimal transport problems. In this pap er\, we address an open problem posed by Bernton et al. and extend their r esults to hold for Schrödinger bridges associated with certain sequences of more general reference measures with enough regularity in a similar sma ll-noise limit. These can be viewed as sequences of entropic optimal trans port plans with non-fixed cost functions. Using a detailed analysis of the associated Skorokhod maps and transition densities\, we show that the new large deviation results cover Schrödinger bridges where the reference pr ocess is a reflected diffusion on bounded convex domains\, corresponding t o recently introduced model choices in the generative modeling literature. \n LOCATION:https://researchseminars.org/talk/gbgstats/103/ END:VEVENT BEGIN:VEVENT SUMMARY:Martin Andrae (Linköping University) DTSTART:20260225T121500Z DTEND:20260225T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/105 DESCRIPTION:Title: Flow-based generative models for data assimilation\nby Martin Andra e (Linköping University) as part of Gothenburg statistics seminar\n\nLect ure held in MVL14.\n\nAbstract\nFlow-based and diffusion generative models have emerged as powerful tools for sampling from complex\, high-dimension al distributions\, such as those found in image generation. In weather for ecasting\, they enable the generation of ensemble forecasts at a fraction of the computational cost of traditional numerical models. These models ha ve also shown promise for solving inverse problems like data assimilation\ , offering advantages over classical methods in high-dimensional\, nonline ar settings. In this talk\, I will introduce the core ideas behind these a pproaches and present some of our recent results.\n LOCATION:https://researchseminars.org/talk/gbgstats/105/ END:VEVENT BEGIN:VEVENT SUMMARY:Filip Rydin (Chalmers\, E2) DTSTART:20260311T121500Z DTEND:20260311T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/106 DESCRIPTION:Title: Learning-based methods for vehicle routing problems - recent advances\nby Filip Rydin (Chalmers\, E2) as part of Gothenburg statistics semina r\n\nLecture held in MVL14.\n\nAbstract\nThis talk reviews recent advances in machine learning for combinatorial optimization\, with a particular fo cus on routing problems such as the Traveling Salesman Problem (TSP) and t he Capacitated Vehicle Routing Problem (CVRP).\n\nFirst\, I will present a unifying high-level hierarchy of methods. I will then delve deeper into e nd-to-end reinforcement learning approaches\, which have shown strong empi rical performance. Finally\, I will present our recent work on multi-objec tive routing over multigraphs\, highlighting how learning-based models can handle competing objectives and complex network structures.\n LOCATION:https://researchseminars.org/talk/gbgstats/106/ END:VEVENT BEGIN:VEVENT SUMMARY:Jun Yang (Department of Mathematical Sciences\, University of Cope nhagen) DTSTART:20260129T121500Z DTEND:20260129T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/109 DESCRIPTION:Title: Stereographic Barker’s MCMC Proposal: Efficiency and Robustness at Yo ur Disposal\nby Jun Yang (Department of Mathematical Sciences\, Univer sity of Copenhagen) as part of Gothenburg statistics seminar\n\nLecture he ld in MVL14.\n\nAbstract\nWe introduce a new family of robust gradient-bas ed MCMC samplers under the framework of stereographic MCMC (Yang et al. 20 22) which maps the original high dimensional problem in Euclidean space on to a sphere. Compared with the existing Stereographic Projection Sampler ( SPS) which is of a random-walk Metropolis type algorithm\, our new family of samplers is gradient-based using the Barker proposal (Livingstone and Z anella\, 2022)\, which improves SPS in high dimensions and is robust to tu ning. Meanwhile\, the proposed algorithms enjoy all the good properties of SPS\, such as uniform ergodicity for a large class of heavy and light-tai led distributions and "blessings of dimensionality".\n LOCATION:https://researchseminars.org/talk/gbgstats/109/ END:VEVENT BEGIN:VEVENT SUMMARY:Easter break DTSTART:20260408T111500Z DTEND:20260408T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/110 DESCRIPTION:by Easter break as part of Gothenburg statistics seminar\n\nLe cture held in MVL14.\nAbstract: TBA\n LOCATION:https://researchseminars.org/talk/gbgstats/110/ END:VEVENT BEGIN:VEVENT SUMMARY:Karl Hammar (Chalmers\, SAAB) DTSTART:20260423T111500Z DTEND:20260423T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/111 DESCRIPTION:Title: Low-Variance Importance Sampling for Discretely Observed Stochastic Dif ferential Equations\nby Karl Hammar (Chalmers\, SAAB) as part of Gothe nburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nStochastic differential equations (SDEs) are commonly used to model dynamical system s of interest. When such systems are observed at discrete times\, they giv e rise to continuous–discrete state space models\, where the main tasks are Bayesian inference of the latent state (filtering and smoothing) as we ll as computation of the observation likelihood (for model comparison). In linear and Gaussian settings these problems can be solved efficiently wit h recursive algorithms like the Kalman filter\, but in nonlinear cases Kal man-type solutions require approximations and lead to biased solutions.\n\ nTo overcome these problems\, alternative solutions include sequential imp ortance sampling (SIS)\, or perhaps most commonly\, sequential importance resampling (SIR)\, also known as particle filters. These methods are unbia sed and converge weakly to the correct solution as the number of particles goes to infinity. However\, the efficiency of these methods depends great ly on the choice of importance distribution\, as poor choices lead to high -variance importance weights and particle degeneracy. The design of good i mportance sampling distributions is therefore of crucial importance. For s moothing\, filtering\, and estimation of the observation likelihood\, the optimal importance distribution is given by the smoothing distribution\, w hich is the focus of this work.\n\nBy Doob’s h-transform\, the smoothing distribution can be characterized as the law of a controlled SDE that dif fers from the unconditional one only by an additional drift term that stee rs trajectories toward future observations. In this work\, we approximate this control term using neural networks\, yielding a tractable approximati on of the smoothing distribution that can be corrected with low-variance i mportance weights. The model is trained using divergence-based objectives\ , including the Kullback–Leibler divergence\, and evaluated in terms of effective sample size and variance of likelihood estimates. This approach reduces variance in importance sampling and improves the efficiency of inf erence in nonlinear SDE models.\n LOCATION:https://researchseminars.org/talk/gbgstats/111/ END:VEVENT BEGIN:VEVENT SUMMARY:Karthik Bharath (University of Nottingham) DTSTART:20260415T111500Z DTEND:20260415T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/113 DESCRIPTION:Title: Rolled Gaussian process models for data as curves on manifolds\nby Karthik Bharath (University of Nottingham) as part of Gothenburg statistic s seminar\n\nLecture held in MVL14.\n\nAbstract\nGiven a planar curve\, im agine rolling a sphere along that curve without slipping or twisting\, and by this means tracing out a curve on the sphere. Such a rolling operation induces a local isometry between the sphere and the plane so that the two curves uniquely determine each other\, and moreover\, the operation exten ds to a general class of manifold $M$ in any dimension $d$. \n\nI will des cribe how rolling can be used to construct an analogue of a Gaussian proce ss with values in $M$\, known as a rolled Gaussian process\, starting from an $\\mathbb R^d$-valued Gaussian process with mean $m$ and covariance $K $. I will discuss the relationship between $m$ and the Frechet mean of the rolled process\, and using the inverse operations of unrolling and unwrap ping\, discuss simple estimators of $m$ and $K$ and their convergence rate s. Utility of the model will be shown in an application involving curves o n 3D orientations coming from a robot learning experiment.\n LOCATION:https://researchseminars.org/talk/gbgstats/113/ END:VEVENT BEGIN:VEVENT SUMMARY:Isac Boström (Chalmers) DTSTART:20260318T121500Z DTEND:20260318T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/114 DESCRIPTION:Title: Half-way seminar: Bayesian Inference for Models of Text Data\nby Is ac Boström (Chalmers) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nModels of text data are increasingly applied to inference tasks in the social sciences to investigate a wide range of l inguistic and cultural phenomena. Word embeddings\, for example\, are comm only used to study semantic change\, political language\, and social bias in large collections of text. However\, these models are typically estimat ed by optimization\, producing point estimates without principled uncertai nty quantification.\n\nIn this talk\, I present a Bayesian formulation of probabilistic word embedding models\, focusing on skip-gram with negative sampling and briefly discussing continuous bag-of-words. I explain why the posterior distribution is non-identifiable under general linear transform ations of the embedding space and introduce a simple and principled constr aint that ensures a well-defined posterior. I then compare different appro aches to posterior inference\, including mean-field variational inference\ , Hamiltonian Monte Carlo\, and Pólya-Gamma Gibbs sampling. By augmenting the likelihood with Pólya-Gamma latent variables\, we obtain an efficien t sampler that provides scalable and well-calibrated uncertainty quantific ation. \n\nI will also briefly discuss the structural topic model as a rel ated example where Bayesian uncertainty plays a central role.\n LOCATION:https://researchseminars.org/talk/gbgstats/114/ END:VEVENT BEGIN:VEVENT SUMMARY:Clara Bertinelli Salucci (University of Oslo) DTSTART:20260429T111500Z DTEND:20260429T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/115 DESCRIPTION:Title: Likelihood Ratio Tests at the Boundary: Asymptotics Beyond Wilks’ the orem\nby Clara Bertinelli Salucci (University of Oslo) as part of Goth enburg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nLikelihoo d ratio tests (LRTs) are a central tool in statistics and data science for hypothesis testing\, typically justified by Wilks’ theorem\, which guar antees an asymptotic chi-square distribution under regular conditions. How ever\, these conditions can fail in practice\, and they do so in a surpris ingly common class of problems: when parameters lie on the boundary of the parameter space. In these settings\, the classical chi-square approximati on is still often used\, but it can be severely misleading\, leading to in correct inference and miscalibrated p-values. This talk introduces the geo metric and probabilistic intuition behind this breakdown and explains how the asymptotic distribution of the LRT changes fundamentally.\n\nI will pr esent a concise derivation of the asymptotic distribution of the LRT under boundary conditions in the case of two parameters\, and interpret it thro ugh the lens of tangent cones and projections of Gaussian random variables . I will then briefly discuss extensions that generalize these results to models with an arbitrary number of parameters on the boundary\, including nuisance parameters.\n\nThe goal of the talk is to provide both theoretica l insight and practical intuition for interpreting LRTs in constrained set tings.\n LOCATION:https://researchseminars.org/talk/gbgstats/115/ END:VEVENT BEGIN:VEVENT SUMMARY:Shuangshuang Chen (KTH) DTSTART:20260506T111500Z DTEND:20260506T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/116 DESCRIPTION:by Shuangshuang Chen (KTH) as part of Gothenburg statistics se minar\n\nLecture held in MVL14.\nAbstract: TBA\n LOCATION:https://researchseminars.org/talk/gbgstats/116/ END:VEVENT BEGIN:VEVENT SUMMARY:Serik Sagitov (Chalmers University of Technology and University of Gothenburg) DTSTART:20260618T111500Z DTEND:20260618T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/117 DESCRIPTION:Title: The Multinomial Allocation Model and the Size of a Randomly Chosen Box< /a>\nby Serik Sagitov (Chalmers University of Technology and University of Gothenburg) as part of Gothenburg statistics seminar\n\nLecture held in M VL14.\n\nAbstract\nThe multinomial allocation model provides a natural fra mework for a generalized birthday problem\, in which n balls are allocated among N boxes with non-uniform allocation probabilities. We revisit a cla ssical asymptotic result due to Kolchin\, Sevastyanov\, and Chistyakov fro m the 1970s\, and reformulate it in terms of the size of a randomly select ed box. Building on this perspective\, we derive a strengthened version of the result by establishing explicit two-sided bounds for the remainder te rms.\n LOCATION:https://researchseminars.org/talk/gbgstats/117/ END:VEVENT BEGIN:VEVENT SUMMARY:Fredrik Sävje (Uppsala University\, Department of Economics) DTSTART:20261028T121500Z DTEND:20261028T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/118 DESCRIPTION:by Fredrik Sävje (Uppsala University\, Department of Economic s) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\nAbs tract: TBA\n LOCATION:https://researchseminars.org/talk/gbgstats/118/ END:VEVENT BEGIN:VEVENT SUMMARY:Måns Magnusson (Uppsala University\, Department of Statistics) DTSTART:20261104T121500Z DTEND:20261104T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/119 DESCRIPTION:by Måns Magnusson (Uppsala University\, Department of Statist ics) as part of Gothenburg statistics seminar\n\nLecture held in MVL14.\nA bstract: TBA\n LOCATION:https://researchseminars.org/talk/gbgstats/119/ END:VEVENT BEGIN:VEVENT SUMMARY:Andrea Zanoni (Scuola Normale Superiore) DTSTART:20260528T111500Z DTEND:20260528T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/121 DESCRIPTION:Title: Learning interaction kernels in stochastic particle systems\nby And rea Zanoni (Scuola Normale Superiore) as part of Gothenburg statistics sem inar\n\nLecture held in MVL14.\n\nAbstract\nInference in stochastic intera cting particle systems is increasingly important due to applications in so cial sciences\, physics\, and machine learning. In this talk\, we focus on learning the interaction kernel from observations of a single particle. W e adopt a semi-parametric approach\, expressing the kernel as a generalize d Fourier series with orthogonal polynomials tailored to the problem. The Fourier coefficients are estimated via a variation of the method of moment s applied to the invariant measure of the mean-field dynamics\, resulting in a linear system based on moments approximated from the particle traject ory. We analyze the approximation error and asymptotic behavior of the est imator in the limits of infinite observation time\, large particle number\ , and increasing number of Fourier coefficients. Numerical experiments ill ustrate the effectiveness of the approach. This work is joint with Grigori os A. Pavliotis (Imperial College London).\n LOCATION:https://researchseminars.org/talk/gbgstats/121/ END:VEVENT BEGIN:VEVENT SUMMARY:Piotr Graczyk (LAREMA Université d'Angers) DTSTART:20260527T111500Z DTEND:20260527T120000Z DTSTAMP:20260422T122728Z UID:gbgstats/122 DESCRIPTION:Title: Penalized estimation for Big Data in Regression Problems and its Geomet ry\nby Piotr Graczyk (LAREMA Université d'Angers) as part of Gothenbu rg statistics seminar\n\nLecture held in MVL14.\n\nAbstract\nI will presen t recent results obtained in [1] and [2] jointly with M. Bogdan\, X. Dupui s\, B. Kolodziejek\, U. Schneider\, T. Skalski\, P. Tardivel and M. Wilczy nski.\n\nIt is well known that LASSO discovers zero coefficients of the ve ctor $b$ in the\nregression equation $Y=Xb+\\varepsilon$ where $X$ is the data matrix and $Y$ the response vector. In fact LASSO estimates the sign of the coefficient vector $b$ ($ b_i$'s positive\, negative or null). The sign is called the model(pattern) of LASSO. In the LASSO estimator the\n$ \\ell^1$ penalty is employed.\n\nIn the study of Big Data one needs to ide ntify more informative patterns of the vector $b$. These leads to use p enalties different from the $\\ell^1$ penalty and to get more dimensional ity reduction. \n\nWe define the pattern of any estimator with polyhedral penalty\, i.e. the unit ball $B$ with respect to the penalty norm is a co nvex polyhedron. Surprising links between the pattern of a penalized estim ator and the geometry of the convex polytope $B^*$ will be explained.\n\n\ n\n We study in detail estimation with a sorted $\\ell^1$ penalty\, calle d SLOPE. Its dual ball $B^*$\n is a signed permutahedron. \n SLOPE is a popular method for dimensionality reduction in the high-dimensional regres sion\, encompassing the LASSO estimator but also the $l^\\infty$ penality . Indeed\, some coefficients of the estimator $\\hat b ^{\\rm SLOPE}$ are null (sparsity) and others are equal in absolute value (clustering). C onsequently\, irrelevant predictors are eliminated and groups of predic tors having the same influence on the\nresponse vector are identified.\nTh e SLOPE pattern of a vector $b$ provides: the sign of its components\, cl usters (components equal in absolute value) and clusters ranking.\n\n In o ur research we give an analytical necessary and sufficient condition for S LOPE pattern recovery of an unknown vector $b$ of regression coefficients. Such condition is called Irrepresentability(IR) condition. For any polyhe dral penalty we find a geometric IR condition.\n\n[1] P. Graczyk\, U. Schn eider\, T. Skalski\, P. Tardivel\, A Unified Framework for Pattern Recove ry in Penalized and Thresholded Estimation and its Geometry\, Journal of Optimization Theory and Applications(2026) vol. 208(1)\, 1-41.\n\n[2] M. Bogdan\, X. Dupuis\, P. Graczyk\, B. Kolodziejek\, T. Skalski\, P. Tardive l\,\nM. Wilczynski\, Pattern recovery by SLOPE\, \nApplied and Computatio nal Harmonic Analysis 80(2026)\, 1-25.\n LOCATION:https://researchseminars.org/talk/gbgstats/122/ END:VEVENT BEGIN:VEVENT SUMMARY:Peter Rudzis (Chalmers) DTSTART:20260325T121500Z DTEND:20260325T130000Z DTSTAMP:20260422T122728Z UID:gbgstats/124 DESCRIPTION:Title: Brownian particle systems with singular interactions\nby Peter Rudz is (Chalmers) as part of Gothenburg statistics seminar\n\nLecture held in MVH12.\n\nAbstract\nWe present a class of Brownian interacting particle sy stems known as \\textit{rank-based diffusions} and their alter ego\, \\tex tit{systems of competing Brownian particles}. The former originally appear ed as a model in stochastic portfolio theory\, while the latter model—ob tained by considering the order statistics of the former—is related to s kew-reflected Brownian motion. This talk will be mainly expository\, descr ibing for a broad audience the fundamental properties of these processes\, including their associated stationary distributions. We will also discuss the infinite-particle versions of these models\, where the stationarity s tructure is richer. As a representative calculation\, we will show that th e distribution of the lowest particle in equilibrium is often Gumbel or re lated. Finally\, we will describe some of our results on the equilibrium f luctuations of a certain space-time random field associated with the infin ite Atlas model (a prototypical model in the class of rank-based diffusion s). These fluctuations have a scaling limit given by a two-parameter Gauss ian process with explicit covariance structure\, equivalently described as the solution to a certain stochastic partial differential equation (SPDE) . As a result\, tagged particles exhibit fluctuations that locally behave as fractional Brownian motion with Hurst parameter 1/4. This work is joint with Sayan Banerjee and Amarjit Budhiraja (UNC Chapel Hill).\n LOCATION:https://researchseminars.org/talk/gbgstats/124/ END:VEVENT END:VCALENDAR