BEGIN:VCALENDAR VERSION:2.0 PRODID:-//jEvents 2.0 for Joomla//EN CALSCALE:GREGORIAN METHOD:PUBLISH BEGIN:VTIMEZONE TZID:America/New_York BEGIN:STANDARD DTSTART:20211116T120000 RDATE:20220313T030000 TZOFFSETFROM:-0400 TZOFFSETTO:-0500 TZNAME:America/New_York EST END:STANDARD BEGIN:STANDARD DTSTART:20221106T010000 RDATE:20230312T030000 TZOFFSETFROM:-0400 TZOFFSETTO:-0500 TZNAME:America/New_York EST END:STANDARD BEGIN:STANDARD DTSTART:20231105T010000 RDATE:20240310T030000 TZOFFSETFROM:-0400 TZOFFSETTO:-0500 TZNAME:America/New_York EST END:STANDARD BEGIN:STANDARD DTSTART:20241103T010000 RDATE:20250309T030000 TZOFFSETFROM:-0400 TZOFFSETTO:-0500 TZNAME:America/New_York EST END:STANDARD BEGIN:STANDARD DTSTART:20251102T010000 RDATE:20260308T030000 TZOFFSETFROM:-0400 TZOFFSETTO:-0500 TZNAME:America/New_York EST END:STANDARD BEGIN:STANDARD DTSTART:20261101T010000 RDATE:20270314T030000 TZOFFSETFROM:-0400 TZOFFSETTO:-0500 TZNAME:America/New_York EST END:STANDARD BEGIN:STANDARD DTSTART:20271107T010000 RDATE:20280312T030000 TZOFFSETFROM:-0400 TZOFFSETTO:-0500 TZNAME:America/New_York EST END:STANDARD BEGIN:STANDARD DTSTART:20281105T010000 RDATE:20290311T030000 TZOFFSETFROM:-0400 TZOFFSETTO:-0500 TZNAME:America/New_York EST END:STANDARD BEGIN:DAYLIGHT DTSTART:20220313T030000 RDATE:20221106T010000 TZOFFSETFROM:-0500 TZOFFSETTO:-0400 TZNAME:America/New_York EDT END:DAYLIGHT BEGIN:DAYLIGHT DTSTART:20230312T030000 RDATE:20231105T010000 TZOFFSETFROM:-0500 TZOFFSETTO:-0400 TZNAME:America/New_York EDT END:DAYLIGHT BEGIN:DAYLIGHT DTSTART:20240310T030000 RDATE:20241103T010000 TZOFFSETFROM:-0500 TZOFFSETTO:-0400 TZNAME:America/New_York EDT END:DAYLIGHT BEGIN:DAYLIGHT DTSTART:20250309T030000 RDATE:20251102T010000 TZOFFSETFROM:-0500 TZOFFSETTO:-0400 TZNAME:America/New_York EDT END:DAYLIGHT BEGIN:DAYLIGHT DTSTART:20260308T030000 RDATE:20261101T010000 TZOFFSETFROM:-0500 TZOFFSETTO:-0400 TZNAME:America/New_York EDT END:DAYLIGHT BEGIN:DAYLIGHT DTSTART:20270314T030000 RDATE:20271107T010000 TZOFFSETFROM:-0500 TZOFFSETTO:-0400 TZNAME:America/New_York EDT END:DAYLIGHT BEGIN:DAYLIGHT DTSTART:20280312T030000 RDATE:20281105T010000 TZOFFSETFROM:-0500 TZOFFSETTO:-0400 TZNAME:America/New_York EDT END:DAYLIGHT END:VTIMEZONE BEGIN:VEVENT UID:6c47582e338dc2475c3bb732c0807906 CATEGORIES:Mathematical Physics Seminar CREATED:20221117T110926 SUMMARY:The unreasonable effectiveness of determinantal processes LOCATION:Hill Center Room 705 DESCRIPTION:In 1960, Wigner published an article famously titled "The Unreasonable Effe ctiveness of Mathematics in the Natural Sciences”. In this talk we will, in a small way, follow the spirit of Wigner’s coinage, and explore the unreas onable effectiveness of determinantal processes (a.k.a. DPPs) far beyond th eir context of origin. DPPs originated in quantum and statistical physics, but have emerged in recent years to be a powerful toolbox for many fundamen tal learning problems. In this talk, we aim to  explore the breadth and dep th of thes applications. On one hand, we will explore a class of Gaussian D PPs and the novel  stochastic geometry of their parameter modulation,and th eir applications to the study of directionality in data and dimension reduc tion. At the other end, we will consider the fundamenta paradigm of stochas tic gradient descent, where we leverage connections with  orthogonal polyno mials to design a minibatch sampling technique based on data-sensitive DPPs ; with provable guarantees for a faster convergence exponent compared to t raditional sampling. Based on the following works.[1] Gaussian determinanta l processes: A new model for directionality in data, with P. Rigollet, Proc eedings of the National Academy of Sciences, vol. 117, no. 24 (2020), pp. 1 3207--13213 (PNAS Direct Submission)[2] Determinantal point processes based on orthogonal polynomials for sampling minibatches in SGD, with R. Bardene t and M. Lin Advances in Neural Information Processing Systems 34 (Spotligh t at NeurIPS 2021)\n X-ALT-DESC;FMTTYPE=text/html:

In 1960, Wigner published an article famously titled "The Unreasonable E ffectiveness of Mathematics in the Natural Sciences”. In this talk we will, in a small way, follow the spirit of Wigner’s coinage, and explore the unr easonable effectiveness of determinantal processes (a.k.a. DPPs) far beyond their context of origin. DPPs originated in quantum and statistical physic s, but have emerged in recent years to be a powerful toolbox for many funda mental learning problems. In this talk, we aim to  explore the breadth and depth of thes applications. On one hand, we will explore a class of Ga ussian DPPs and the novel  stochastic geometry of their parameter modu lation,and their applications to the study of directionality in data and di mension reduction. At the other end, we will consider the fundamenta paradi gm of stochastic gradient descent, where we leverage connections with  orthogonal polynomials to design a minibatch sampling technique based on d ata-sensitive DPPs ; with provable guarantees for a faster convergence expo nent compared to traditional sampling. Based on the following works.[1] Gau ssian determinantal processes: A new model for directionality in data, with P. Rigollet, Proceedings of the National Academy of Sciences, vol. 117, no . 24 (2020), pp. 13207--13213 (PNAS Direct Submission)[2] Determinantal poi nt processes based on orthogonal polynomials for sampling minibatches in SG D, with R. Bardenet and M. Lin Advances in Neural Information Processing Sy stems 34 (Spotlight at NeurIPS 2021)

CONTACT:Subhro Ghosh - National University of Singapore DTSTAMP:20260830T041904 DTSTART;TZID=America/New_York:20221117T120000 DTEND;TZID=America/New_York:20221117T130000 SEQUENCE:0 TRANSP:OPAQUE END:VEVENT END:VCALENDAR