BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CMSA - ECPv6.17.1//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:CMSA
X-ORIGINAL-URL:https://cmsa.fas.harvard.edu
X-WR-CALDESC:Events for CMSA
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20190310T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20191103T060000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20200308T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20201101T060000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:20210314T070000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:20211107T060000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200624T093000
DTEND;TZID=America/New_York:20200624T120000
DTSTAMP:20240216T103102Z
CREATED:20240216T103102Z
LAST-MODIFIED:20240216T103102Z
UID:10002765-1592991000-1593000000@cmsa.fas.harvard.edu
SUMMARY:6/24/2020 Quantum Matter Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/6-24-2020-quantum-matter-seminar-2/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200624T103000
DTEND;TZID=America/New_York:20200624T120000
DTSTAMP:20240216T074231Z
CREATED:20240209T111631Z
LAST-MODIFIED:20240216T074231Z
UID:10001849-1592994600-1593000000@cmsa.fas.harvard.edu
SUMMARY:6/24/2020 Quantum Matter Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/6-24-2020-quantum-matter-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200625T093000
DTEND;TZID=America/New_York:20200625T110000
DTSTAMP:20240209T112140Z
CREATED:20240209T112140Z
LAST-MODIFIED:20240209T112140Z
UID:10001853-1593077400-1593082800@cmsa.fas.harvard.edu
SUMMARY:6/25/2020 Condensed Matter Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/6-25-2020-condensed-matter-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200707T093000
DTEND;TZID=America/New_York:20200707T223000
DTSTAMP:20240209T110534Z
CREATED:20240209T110534Z
LAST-MODIFIED:20240209T110534Z
UID:10001843-1594114200-1594161000@cmsa.fas.harvard.edu
SUMMARY:7/7/2020 Geometry and Physics Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/7-7-2020-geometry-and-physics-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200709T103000
DTEND;TZID=America/New_York:20200709T120000
DTSTAMP:20240209T111905Z
CREATED:20240209T111905Z
LAST-MODIFIED:20240209T111905Z
UID:10001851-1594290600-1594296000@cmsa.fas.harvard.edu
SUMMARY:7/9/2020 Condensed Matter Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/7-9-2020-condensed-matter-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200713T100000
DTEND;TZID=America/New_York:20200713T110000
DTSTAMP:20240209T014801Z
CREATED:20240209T014801Z
LAST-MODIFIED:20240209T014801Z
UID:10001783-1594634400-1594638000@cmsa.fas.harvard.edu
SUMMARY:Social Science Applications Forum
DESCRIPTION:During the Summer of 2020\, the CMSA will be hosting a periodic Social Science Applications Seminar. \nThe list of speakers is below and will be updated as details are confirmed. \nFor a list of past Social Science Applications talks\, please click here.\n\n\n\n\nDate\nSpeaker\nTitle/Abstract\n\n\n\n\n7/13/2020 10:00-11:00am ET\nLudovic Tangpi (Princeton)\nPlease note\, this seminar will take place online using Zoom. \nTitle: Convergence of Large Population Games to Mean Field Games with Interaction Through the Controls \nAbstract: This work considers stochastic differential games with a large number of players\, whose costs and dynamics interact through the empirical distribution of both their states and their controls. We develop a framework to prove convergence of finite-player games to the asymptotic mean field game. Our approach is based on the concept of propagation of chaos for forward and backward weakly interacting particles which we investigate by fully probabilistic methods\, and which appear to be of independent interest. These propagation of chaos arguments allow to derive moment and concentration bounds for the convergence of both Nash equilibria and social optima in non-cooperative and cooperative games\, respectively. Incidentally\, we also obtain convergence of a system of second order parabolic partial differential equations on finite dimensional spaces to a second order parabolic partial differential equation on the Wasserstein space.\nFor security reasons\, you will have to show your full name to join the meeting.\n\n\n7/27/2020\n10:00pm\nMichael Ewens (Caltech)\nPlease note\, this seminar will take place online using Zoom. \nTitle: Measuring Intangible Capital with Market Prices \nAbstract: Despite the importance of intangibles in today’s economy\, current standards prohibit the capitalization of internally created knowledge and organizational capital\, resulting in a downward bias of reported assets. As a result\, researchers estimate this value by capitalizing prior flows of R&D and SG&A. In doing so\, a set of capitalization parameters\, i.e. the R&D depreciation rate and the fraction of SG&A that represents a long-lived asset\, must be assumed. Parameters now in use are derived from models with strong assumptions or are ad hoc. We develop a capitalization model that motivates the use of market prices of intangibles to estimate these parameters. Two settings provide intangible asset values: (1) publicly traded equity prices and (2) acquisition prices. We use these parameters to estimate intangible capital stocks and subject them to an extensive set of diagnostic analyses that compare them with stocks estimated using existing parameters. Intangible stocks developed from exit price parameters outperform both stocks developed by publicly traded parameters and those stocks developed with existing estimates. (Joint work with Ryan Peters and Sean Wang.)
URL:https://cmsa.fas.harvard.edu/event/social-science-applications-forum/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200713T213000
DTEND;TZID=America/New_York:20200713T223000
DTSTAMP:20240216T083015Z
CREATED:20240209T105819Z
LAST-MODIFIED:20240216T083015Z
UID:10001839-1594675800-1594679400@cmsa.fas.harvard.edu
SUMMARY:7/13/2020 Geometry and Physics Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/7-13-2020-geometry-and-physics-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200715T103000
DTEND;TZID=America/New_York:20200715T120000
DTSTAMP:20240209T110815Z
CREATED:20240209T110815Z
LAST-MODIFIED:20240209T110815Z
UID:10001845-1594809000-1594814400@cmsa.fas.harvard.edu
SUMMARY:7/15/2020 Quantum Matter Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/7-15-2020-quantum-matter-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200716T103000
DTEND;TZID=America/New_York:20200716T120000
DTSTAMP:20240209T111248Z
CREATED:20240209T111248Z
LAST-MODIFIED:20240209T111248Z
UID:10001847-1594895400-1594900800@cmsa.fas.harvard.edu
SUMMARY:7/16/2020 Condensed Matter Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/7-16-2020-condensed-matter-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200720T093000
DTEND;TZID=America/New_York:20200720T103000
DTSTAMP:20240209T155712Z
CREATED:20240209T155712Z
LAST-MODIFIED:20240209T155712Z
UID:10001873-1595237400-1595241000@cmsa.fas.harvard.edu
SUMMARY:7/20/2020 Geometry and Physics Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/7-20-2020-geometry-and-physics-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200723T093000
DTEND;TZID=America/New_York:20200723T110000
DTSTAMP:20240209T105853Z
CREATED:20240209T105122Z
LAST-MODIFIED:20240209T105853Z
UID:10001833-1595496600-1595502000@cmsa.fas.harvard.edu
SUMMARY:7/23/2020 Quantum Matter Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/7-23-2020-quantum-matter-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200729T153000
DTEND;TZID=America/New_York:20200729T170000
DTSTAMP:20240209T110242Z
CREATED:20240209T110242Z
LAST-MODIFIED:20240209T110242Z
UID:10001842-1596036600-1596042000@cmsa.fas.harvard.edu
SUMMARY:7/29/2020 Quantum Matter Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/7-29-2020-quantum-matter-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200730T093000
DTEND;TZID=America/New_York:20200730T110000
DTSTAMP:20240216T092740Z
CREATED:20240216T092740Z
LAST-MODIFIED:20240216T092740Z
UID:10002759-1596101400-1596106800@cmsa.fas.harvard.edu
SUMMARY:7/30/2020 Condensed Matters Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/7-30-2020-condensed-matters-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200731T150000
DTEND;TZID=America/New_York:20200731T163000
DTSTAMP:20240216T093126Z
CREATED:20240216T093126Z
LAST-MODIFIED:20240216T093126Z
UID:10002761-1596207600-1596213000@cmsa.fas.harvard.edu
SUMMARY:7/31/2020 Quantum Matter Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/7-31-2020-quantum-matter-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200804T093000
DTEND;TZID=America/New_York:20200804T103000
DTSTAMP:20240216T092559Z
CREATED:20240216T092559Z
LAST-MODIFIED:20240216T092559Z
UID:10002758-1596533400-1596537000@cmsa.fas.harvard.edu
SUMMARY:8/4/2020 Geometry and Physics Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/8-4-2020-geometry-and-physics-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200811T093000
DTEND;TZID=America/New_York:20200811T103000
DTSTAMP:20240216T092323Z
CREATED:20240216T092323Z
LAST-MODIFIED:20240216T092323Z
UID:10002757-1597138200-1597141800@cmsa.fas.harvard.edu
SUMMARY:8/11/2020 Geometry and Physics Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/8-11-2020-geometry-and-physics-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200818T093000
DTEND;TZID=America/New_York:20200818T113000
DTSTAMP:20240216T091632Z
CREATED:20240216T091507Z
LAST-MODIFIED:20240216T091632Z
UID:10002756-1597743000-1597750200@cmsa.fas.harvard.edu
SUMMARY:8/18/2020 Geometry and Physics Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/8-18-2020-geometry-and-physics-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200819T093000
DTEND;TZID=America/New_York:20200819T110000
DTSTAMP:20240216T090831Z
CREATED:20240216T090831Z
LAST-MODIFIED:20240216T090831Z
UID:10002753-1597829400-1597834800@cmsa.fas.harvard.edu
SUMMARY:8/19/2020 Quantum Matter Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/8-19-2020-quantum-matter-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200820T093000
DTEND;TZID=America/New_York:20200820T110000
DTSTAMP:20240216T091121Z
CREATED:20240216T091121Z
LAST-MODIFIED:20240216T091121Z
UID:10002754-1597915800-1597921200@cmsa.fas.harvard.edu
SUMMARY:8/20/2020 Quantum Matter
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/8-20-2020-quantum-matter/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200825T093000
DTEND;TZID=America/New_York:20200825T103000
DTSTAMP:20240209T023148Z
CREATED:20240209T023148Z
LAST-MODIFIED:20240209T023148Z
UID:10001808-1598347800-1598351400@cmsa.fas.harvard.edu
SUMMARY:8/25/2020 Geometry and Physics Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/8-25-2020-geometry-and-physics-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200826T093000
DTEND;TZID=America/New_York:20200826T110000
DTSTAMP:20240712T163525Z
CREATED:20240216T090201Z
LAST-MODIFIED:20240712T163525Z
UID:10002751-1598434200-1598439600@cmsa.fas.harvard.edu
SUMMARY:8/26/2020 Quantum Matter Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/8-26-2020-quantum-matter-seminar/
LOCATION:MA
CATEGORIES:Quantum Matter,Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200827T093000
DTEND;TZID=America/New_York:20200827T110000
DTSTAMP:20240209T023300Z
CREATED:20240209T023300Z
LAST-MODIFIED:20240209T023300Z
UID:10001809-1598520600-1598526000@cmsa.fas.harvard.edu
SUMMARY:8/27/2020 Quantum Matter Seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/8-27-2020-quantum-matter-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200916T150000
DTEND;TZID=America/New_York:20200916T160000
DTSTAMP:20240515T183741Z
CREATED:20240209T021453Z
LAST-MODIFIED:20240515T183741Z
UID:10001799-1600268400-1600272000@cmsa.fas.harvard.edu
SUMMARY:Graph Representation Learning: Recent Advances and Open Challenges
DESCRIPTION:Speaker: William Hamilton\, McGill University and MILA \nTitle: Graph Representation Learning: Recent Advances and Open Challenges \nAbstract: Graph-structured data is ubiquitous throughout the natural and social sciences\, from telecommunication networks to quantum chemistry. Building relational inductive biases into deep learning architectures is crucial if we want systems that can learn\, reason\, and generalize from this kind of data. Recent years have seen a surge in research on graph representation learning\, most prominently in the development of graph neural networks (GNNs). Advances in GNNs have led to state-of-the-art results in numerous domains\, including chemical synthesis\, 3D-vision\, recommender systems\, question answering\, and social network analysis. In the first part of this talk I will provide an overview and summary of recent progress in this fast-growing area\, highlighting foundational methods and theoretical motivations. In the second part of this talk I will discuss fundamental limitations of the current GNN paradigm and propose open challenges for the theoretical advancement of the field. \n 
URL:https://cmsa.fas.harvard.edu/event/9-16-2020-new-technologies-seminar/
LOCATION:Virtual
CATEGORIES:New Technologies in Mathematics Seminar
ATTACH;FMTTYPE=image/png:https://cmsa.fas.harvard.edu/media/CMSA-New-Technologies-in-Mathematics-09.16.20-1.png
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20200923T150000
DTEND;TZID=America/New_York:20200923T160000
DTSTAMP:20240515T201416Z
CREATED:20240209T014918Z
LAST-MODIFIED:20240515T201416Z
UID:10001784-1600873200-1600876800@cmsa.fas.harvard.edu
SUMMARY:Self-induced regularization from linear regression to neural networks
DESCRIPTION:Speaker: Andrea Montanari\, Departments of Electrical Engineering and Statistics\, Stanford \nTitle: Self-induced regularization from linear regression to neural networks \nAbstract: Modern machine learning methods –most noticeably multi-layer neural networks– require to fit highly non-linear models comprising tens of thousands to millions of parameters. Despite this\, little attention is paid to the regularization mechanism to control model’s complexity. Indeed\, the resulting models are often so complex as to achieve vanishing training error: they interpolate the data. Despite this\, these models generalize well to unseen data : they have small test error. I will discuss several examples of this phenomenon\, beginning with a simple linear regression model\, and ending with two-layers neural networks in the so-called lazy regime. For these examples precise asymptotics could be determined mathematically\, using tools from random matrix theory. I will try to extract a unifying picture. A common feature is the fact that a complex unregularized nonlinear model becomes essentially equivalent to a simpler model\, which is however regularized in a non-trivial way. [Based on joint papers with: Behrooz Ghorbani\, Song Mei\, Theodor Misiakiewicz\, Feng Ruan\, Youngtak Sohn\, Jun Yan\, Yiqiao Zhong] \n  \n 
URL:https://cmsa.fas.harvard.edu/event/9-23-2020-new-tech-in-mathematics-seminar/
LOCATION:Virtual
CATEGORIES:New Technologies in Mathematics Seminar
ATTACH;FMTTYPE=image/png:https://cmsa.fas.harvard.edu/media/CMSA-New-Technologies-in-Mathematics-09.23.20.png
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20201001T103000
DTEND;TZID=America/New_York:20201001T120000
DTSTAMP:20240209T013427Z
CREATED:20240209T013427Z
LAST-MODIFIED:20240209T013427Z
UID:10001774-1601548200-1601553600@cmsa.fas.harvard.edu
SUMMARY:10/1/2020 Quantum Matter seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/10-1-2020-quantum-matter-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20201005T103000
DTEND;TZID=America/New_York:20201005T120000
DTSTAMP:20240209T104813Z
CREATED:20240209T104813Z
LAST-MODIFIED:20240209T104813Z
UID:10001831-1601893800-1601899200@cmsa.fas.harvard.edu
SUMMARY:4/15/2020 Quantum Matter seminar
DESCRIPTION:
URL:https://cmsa.fas.harvard.edu/event/4-15-2020-quantum-matter-seminar/
LOCATION:MA
CATEGORIES:Seminars
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20201014T090000
DTEND;TZID=America/New_York:20201014T100000
DTSTAMP:20240507T194446Z
CREATED:20240127T031011Z
LAST-MODIFIED:20240507T194446Z
UID:10001499-1602666000-1602669600@cmsa.fas.harvard.edu
SUMMARY:Statistical\, mathematical\, and computational aspects of noisy intermediate-scale quantum computers 
DESCRIPTION:Speaker: Gil Kalai (Hebrew University and IDC Herzliya) \nTitle: Statistical\, mathematical\, and computational aspects of noisy intermediate-scale quantum computers \nAbstract: Noisy intermediate-scale quantum (NISQ) Computers hold the key for important theoretical and experimental questions regarding quantum computers. In the lecture I will describe some questions about mathematics\, statistics and computational complexity which arose in my study of NISQ systems and are related to \n\na) My general argument “against” quantum computers\,\nb) My analysis (with Yosi Rinott and Tomer Shoham) of the Google 2019 “quantum supremacy” experiment.\nRelevant papers:\nYosef Rinott\, Tomer Shoham and Gil Kalai\, Statistical aspects of the quantum supremacy demonstration\, https://gilkalai.files.wordpress.com/2019/11/stat-quantum2.pdf\nGil Kalai\, The Argument against Quantum Computers\, the Quantum Laws of Nature\, and Google’s Supremacy Claims\, https://gilkalai.files.wordpress.com/2020/08/laws-blog2.pdf\nGil Kalai\, Three puzzles on mathematics\, computations\, and games\, https://gilkalai.files.wordpress.com/2019/09/main-pr.pdf
URL:https://cmsa.fas.harvard.edu/event/10-14-2020-colloquium/
LOCATION:MA
CATEGORIES:Colloquium
ATTACH;FMTTYPE=image/png:https://cmsa.fas.harvard.edu/media/CMSA-Colloquium-10.14.20-1.png
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20201014T150000
DTEND;TZID=America/New_York:20201014T160000
DTSTAMP:20240515T192014Z
CREATED:20240201T021720Z
LAST-MODIFIED:20240515T192014Z
UID:10001522-1602687600-1602691200@cmsa.fas.harvard.edu
SUMMARY:Triple Descent and a Fine-Grained Bias-Variance Decomposition
DESCRIPTION:Speaker: Jeffrey Pennington\, Google Brain \nTitle: Triple Descent and a Fine-Grained Bias-Variance Decomposition \nAbstract: Classical learning theory suggests that the optimal generalization performance of a machine learning model should occur at an intermediate model complexity\, striking a balance between simpler models that exhibit high bias and more complex models that exhibit high variance of the predictive function. However\, such a simple trade-off does not adequately describe the behavior of many modern deep learning models\, which simultaneously attain low bias and low variance in the heavily overparameterized regime. Recent efforts to explain this phenomenon theoretically have focused on simple settings\, such as linear regression or kernel regression with unstructured random features\, which are too coarse to reveal important nuances of actual neural networks. In this talk\, I will describe a precise high-dimensional asymptotic analysis of Neural Tangent Kernel regression that reveals some of these nuances\, including non-monotonic behavior deep in the overparameterized regime. I will also present a novel bias-variance decomposition that unambiguously attributes these surprising observations to particular sources of randomness in the training procedure.
URL:https://cmsa.fas.harvard.edu/event/10-14-2020-new-technologies-seminar/
LOCATION:MA
CATEGORIES:New Technologies in Mathematics Seminar
ATTACH;FMTTYPE=image/png:https://cmsa.fas.harvard.edu/media/CMSA-New-Technologies-in-Mathematics-10.14.20.png
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20201028T150000
DTEND;TZID=America/New_York:20201028T160000
DTSTAMP:20240515T201157Z
CREATED:20240127T031359Z
LAST-MODIFIED:20240515T201157Z
UID:10001502-1603897200-1603900800@cmsa.fas.harvard.edu
SUMMARY:Generalization bounds for rational self-supervised learning algorithms\, or "Understanding generalizations requires rethinking deep learning"
DESCRIPTION:Speakers: Boaz Barak and Yamini Bansal\, Harvard University Dept. of Computer Science \nTitle: Generalization bounds for rational self-supervised learning algorithms\, or “Understanding generalizations requires rethinking deep learning” \nAbstract: The generalization gap of a learning algorithm is the expected difference between its performance on the training data and its performance on fresh unseen test samples. Modern deep learning algorithms typically have large generalization gaps\, as they use more parameters than the size of their training set. Moreover the best known rigorous bounds on their generalization gap are often vacuous. In this talk we will see a new upper bound on the generalization gap of classifiers that are obtained by first using self-supervision to learn a complex representation of the (label free) training data\, and then fitting a simple (e.g.\, linear) classifier to the labels. Such classifiers have become increasingly popular in recent years\, as they offer several practical advantages and have been shown to approach state-of-art results. We show that (under the assumptions described below) the generalization gap of such classifiers tends to zero as long as the complexity of the simple classifier is asymptotically smaller than the number of training samples. We stress that our bound is independent of the complexity of the representation that can use an arbitrarily large number of parameters. Our bound assuming that the learning algorithm satisfies certain noise-robustness (adding small amount of label noise causes small degradation in performance) and rationality (getting the wrong label is not better than getting no label at all) conditions that widely (and sometimes provably) hold across many standard architectures. We complement this result with an empirical study\, demonstrating that our bound is non-vacuous for many popular representation-learning based classifiers on CIFAR-10 and ImageNet\, including SimCLR\, AMDIM and BigBiGAN. The talk will not assume any specific background in machine learning\, and should be accessible to a general mathematical audience. Joint work with Gal Kaplun. \n 
URL:https://cmsa.fas.harvard.edu/event/10-28-2020-new-technologies-in-mathematics-seminar/
LOCATION:MA
CATEGORIES:New Technologies in Mathematics Seminar
ATTACH;FMTTYPE=image/png:https://cmsa.fas.harvard.edu/media/CMSA-New-Technologies-in-Mathematics-10.28.20.png
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/New_York:20201104T150000
DTEND;TZID=America/New_York:20201104T160000
DTSTAMP:20240515T200835Z
CREATED:20240127T021940Z
LAST-MODIFIED:20240515T200835Z
UID:10001486-1604502000-1604505600@cmsa.fas.harvard.edu
SUMMARY:Some exactly solvable models for machine learning via Statistical physics
DESCRIPTION:Speaker: Florent Krzakala\, EPFL \nTitle: Some exactly solvable models for machine learning via Statistical physics \nAbstract: The increasing dimensionality of data in the modern machine learning age presents new challenges and opportunities. The high dimensional settings allow one to use powerful asymptotic methods from probability theory and statistical physics to obtain precise characterizations and develop new algorithmic approaches. Statistical mechanics approaches\, in particular\, are very well suited for such problems. Will give examples of recent works in our group that build on powerful methods of statistical physics of disordered systems to analyze some relevant questions in machine learning and neural networks\, including overparameterization\, kernel methods\, and the behavior gradient descent algorithm in a high dimensional non-convex landscape.
URL:https://cmsa.fas.harvard.edu/event/11-4-2020-new-technologies-in-math/
LOCATION:Virtual
CATEGORIES:New Technologies in Mathematics Seminar
ATTACH;FMTTYPE=image/png:https://cmsa.fas.harvard.edu/media/CMSA-New-Technologies-in-Mathematics-11.04.20.png
END:VEVENT
END:VCALENDAR