BEGIN:VCALENDAR VERSION:2.0 PRODID:researchseminars.org CALSCALE:GREGORIAN X-WR-CALNAME:researchseminars.org BEGIN:VEVENT SUMMARY:Rui Castro (Mathematics Department\, TU Eindhoven) DTSTART:20230518T160000Z DTEND:20230518T170000Z DTSTAMP:20260423T021049Z UID:MPML/107 DESCRIPTION:Title: A nomaly detection for a large number of streams: a permutation/rank-based h igher criticism approach\nby Rui Castro (Mathematics Department\, TU E indhoven) as part of Mathematics\, Physics and Machine Learning (IST\, Lis bon)\n\n\nAbstract\nAnomaly detection when observing a large number of dat a streams is essential in a variety of applications\, ranging from epidemi ological studies to monitoring of complex systems. High-dimensional scenar ios are usually tackled with scan-statistics and related methods\, requiri ng stringent distributional assumptions for proper test calibration. In th is talk we take a non-parametric stance\, and introduce two variants of th e higher criticism test that do not require knowledge of the null distribu tion for proper calibration. In the first variant we calibrate the test by permutation\, while in the second variant we use a rank-based approach. B oth methodologies result in exact tests in finite samples. Our permutation methodology is applicable when observations within null streams are indep endent and identically distributed\, and we show this methodology is asymp totically optimal in the wide class of exponential models. Our rank-based methodology is more flexible\, and only requires observations within null streams to be independent. We provide an asymptotic characterization of th e power of the test in terms of the probability of mis-ranking null observ ations\, showing that the asymptotic power loss (relative to an oracle tes t) is minimal for many common models. As the proposed statistics do not re ly on asymptotic approximations\, they typically perform better than popul ar variants of higher criticism relying on such approximations. Finally\, we demonstrate the use of these methodologies when monitoring the content uniformity of an active ingredient for a batch-produced drug product\, and monitoring the daily number of COVID-19 cases in the Netherlands.\n\nBase d on joint work with Ivo Stoepker\, Ery Arias-Castro and Edwin van de den Heuvel:\nhttps://arxiv.org/abs/2009.03117\n LOCATION:https://researchseminars.org/talk/MPML/107/ END:VEVENT END:VCALENDAR