BEGIN:VCALENDAR VERSION:2.0 PRODID:researchseminars.org CALSCALE:GREGORIAN X-WR-CALNAME:researchseminars.org BEGIN:VEVENT SUMMARY:Francesca Zaffora Blando (Carnegie Mellon University) DTSTART:20211206T213000Z DTEND:20211206T223000Z DTSTAMP:20260423T005743Z UID:CTA/68 DESCRIPTION:Title: Alg orithmic randomness and Bayesian convergence\nby Francesca Zaffora Bla ndo (Carnegie Mellon University) as part of Computability theory and appli cations\n\n\nAbstract\nMuch recent work in algorithmic randomness has conc erned\ncharacterizations of randomness notions in terms of effectivization s of\nalmost-everywhere convergence theorems in analysis and probability t heory. In\nthis talk\, I will consider some results that are part of the b asic toolkit of\nBayesian epistemologists from this perspective. In partic ular\, I will focus on\ncertain martingale convergence theorems that form one of the cornerstones of\nBayesian epistemology and that fall under the general umbrella of "Bayesian\nconvergence-to-the-truth results". These re sults are standardly taken to\nestablish that a Bayesian agent’s beliefs are guaranteed to converge to the\ntruth with probability one as the evid ence accumulates. We will see that\, for\ncomputable Bayesian agents (i.e. \, Bayesian agents with computable priors)\, we\nnot only have that conver gence to the truth occurs with probability one\, but\nwe can also provide precise characterizations of the data streams along which\nbeliefs converg e to the truth: they are precisely the algorithmically random\ndata stream s. I will conclude by sketching a broader computability-theoretic\napproac h to Bayesian epistemology suggested by these results.\n LOCATION:https://researchseminars.org/talk/CTA/68/ END:VEVENT END:VCALENDAR