BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Halıcıoğlu Data Science Institute - UC San Diego - ECPv6.16.2//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-ORIGINAL-URL:https://datascience.ucsd.edu
X-WR-CALDESC:Events for Halıcıoğlu Data Science Institute - UC San Diego
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:America/Los_Angeles
BEGIN:DAYLIGHT
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
DTSTART:20190310T100000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
DTSTART:20191103T090000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
DTSTART:20200308T100000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
DTSTART:20201101T090000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
DTSTART:20210314T100000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
DTSTART:20211107T090000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
DTSTART:20220313T100000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
DTSTART:20221106T090000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
DTSTART:20230312T100000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
DTSTART:20231105T090000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
DTSTART:20240310T100000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
DTSTART:20241103T090000
END:STANDARD
BEGIN:DAYLIGHT
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
DTSTART:20250309T100000
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
DTSTART:20251102T090000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=America/Los_Angeles:20240213T130000
DTEND;TZID=America/Los_Angeles:20240213T140000
DTSTAMP:20260530T052804
CREATED:20240209T164305Z
LAST-MODIFIED:20240209T164305Z
UID:10000440-1707829200-1707832800@datascience.ucsd.edu
SUMMARY:On Data Ecology\, Data Markets\, the Value of Data\, and Dataflow Governance | Raul Castro Fernandez
DESCRIPTION:Abstract:\nData shapes our social\, economic\, cultural\, and technological environments. Data is valuable\, so people seek it\, inducing data to flow. The resulting dataflows distribute data and thus value. For example\, large Internet companies profit from accessing data from their users\, and engineers of large language models seek large and diverse data sources to train powerful models. It is possible to judge the impact of data in an environment by analyzing how the dataflows in that environment impact the participating agents. My research hypothesizes that it is also possible to design (better) data environments by controlling what dataflows materialize; not only can we analyze environments but also synthesize them. In this talk\, I present the research agenda on “data ecology\,” which seeks to build the principles\, theory\, algorithms\, and systems to design beneficial data environments. I will also present examples of data environments my group has designed\, including data markets for machine learning\, data-sharing\, and data integration. I will conclude by discussing the impact of dataflows in data governance and how the ideas are interwoven with the concepts of trust\, privacy\, and the elusive notion of “data value.” As part of the technical discussion\, I will complement the data market designs with the design of a data escrow system that permits controlling dataflows.\nBio:\nIn my research\, I ask what is the value of data and explore the potential of data markets to unlock that value. My group collaborates with economists\, legal scholars\, statisticians\, and domain scientists. We build systems to share\, discover\, prepare\, integrate\, and process data. I have traditionally worked on distributed query processing systems and continue to do so. I have received a SIGMOD’23 Test-of-time-Award. I am an assistant professor in the Department of Computer Science and on the Committee of Data Science at The University of Chicago. Before UChicago\, I did a postdoc at MIT with Sam Madden and Mike Stonebraker. And before that\, I completed a PhD at Imperial College London with Peter Pietzuch.
URL:https://datascience.ucsd.edu/event/on-data-ecology-data-markets-the-value-of-data-and-dataflow-governance-raul-castro-fernandez/
CATEGORIES:Colloquium,Seminar
ATTACH;FMTTYPE=image/png:https://datascience.ucsd.edu/wp-content/uploads/2024/01/HDSI-UCSD-Image-e1712856546428.png
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/Los_Angeles:20240209T130000
DTEND;TZID=America/Los_Angeles:20240209T140000
DTSTAMP:20260530T052804
CREATED:20240209T164509Z
LAST-MODIFIED:20240209T164509Z
UID:10000439-1707483600-1707487200@datascience.ucsd.edu
SUMMARY:EnCORE : Theoretical Exploration of Foundation Model Adaptation\, Kangwook Lee\, UW Madison\, Feb 9th\, 1-2pm
DESCRIPTION:Abstract: Due to the enormous size of foundation models\, various new methods for efficient model adaptation have been developed. Parameter-efficient fine-tuning (PEFT) is an adaptation method that updates only a tiny fraction of the model parameters\, leaving the remainder unchanged. In-context Learning (ICL) is a test-time adaptation method\, which repurposes foundation models by providing them with labeled samples as part of the input context. Given the growing importance of this emerging paradigm\, developing theoretical foundations for the new paradigm is of utmost importance. \nIn this talk\, I will introduce two preliminary results toward this goal. In the first part\, I will present a theoretical analysis of Low-Rank Adaptation (also known as LoRA)\, one of the most popular PEFT methods today. Our analysis of the expressive power of LoRA not only helps us better understand the high adaptivity of LoRA observed in practice but also provides insights to practitioners. In the second part\, I will introduce our probabilistic framework for a better understanding of ICL. With our framework\, one can analyze the transition between two distinct modes of ICL: task retrieval and learning. We also discuss how our framework can help explain and predict various phenomena\, which can be observed with large language models in practice yet not fully explained. \nBio: Kangwook Lee is an Assistant Professor in the Electrical and Computer Engineering Department and the Computer Sciences Department (by courtesy) at the University of Wisconsin-Madison. Previously\, he was a Research Assistant Professor at the Information and Electronics Research Institute of KAIST and was a postdoctoral scholar at the same institute. He received his PhD in 2016 from the Electrical Engineering and Computer Science department at UC Berkeley. He is the recipient of the IEEE Joint Communications Society/Information Theory Society Paper Award (2020) and the KSEA Young Investigator Grant Award (2022).
URL:https://datascience.ucsd.edu/event/encore-theoretical-exploration-of-foundation-model-adaptation-kangwook-lee-uw-madison-feb-9th-1-2pm/
LOCATION:Atkinson Hall\, Fourth Floor
CATEGORIES:Colloquium,Seminar
ATTACH;FMTTYPE=image/png:https://datascience.ucsd.edu/wp-content/uploads/2023/10/Encore-logo_HDSI-Website.png
ORGANIZER;CN="k1omerry@ucsd.edu":MAILTO:k1omerry@ucsd.edu
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/Los_Angeles:20230609T150000
DTEND;TZID=America/Los_Angeles:20230609T160000
DTSTAMP:20260530T052804
CREATED:20230605T233443Z
LAST-MODIFIED:20230914T164123Z
UID:10000392-1686322800-1686326400@datascience.ucsd.edu
SUMMARY:Deep Latent Variable Models for Compression and Natural Science | Stephan Mandt
DESCRIPTION:Abstract: Latent variable models have been an integral part of probabilistic machine learning\, ranging from simple mixture models to variational autoencoders to powerful diffusion probabilistic models at the center of recent media attention. Perhaps less well-appreciated is the intimate connection between latent variable models and data compression\, and the potential of these models for advancing natural science. This talk will explore these topics. I will begin by showcasing connections between variational methods and the theory and practice of neural data compression. On the applied side\, variational methods lead to machine-learned compressors of data such as images and videos and offer principled techniques for enhancing their compression performance\, as well as reducing their decoding complexity. On the theory side\, variational methods also provide scalable bounds on the fundamental compressibility of real-world data\, such as images and particle physics data. Lastly\, I will also delve into climate science projects\, where a combination of deep latent variable modeling and vector quantization enables assessing distribution shifts induced by varying climate models and the effects of global warming. \nBio: Stephan Mandt is an Associate Professor of Computer Science and Statistics at the University of California\, Irvine. From 2016 until 2018\, he was a Senior Researcher and Head of the statistical machine learning group at Disney Research in Pittsburgh and Los Angeles. He held previous postdoctoral positions at Columbia University and Princeton University. Stephan holds a Ph.D. in Theoretical Physics from the University of Cologne in Germany\, where he received the National Merit Scholarship. He received the NSF CAREER Award\, a Kavli Fellowship of the U.S. National Academy of Sciences\, the German Research Foundation’s Mercator Fellowship\, and the UCI ICS Mid-Career Excellence in Research Award. He is a member of the ELLIS Society and a former visiting researcher at Google Brain. Stephan will serve as Program Chair of the AISTATS 2024 conference\, currently serves as an Action Editor for JMLR and TMLR\, and frequently serves as Area Chair for NeurIPS\, ICML\, AAAI\, and ICLR.
URL:https://datascience.ucsd.edu/event/deep-latent-variable-models-for-compression-and-natural-science-stephan-mandt/
LOCATION:Computer Science & Engineering Building (CSE)\, Room 1202
CATEGORIES:Colloquium
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/Los_Angeles:20230531T140000
DTEND;TZID=America/Los_Angeles:20230531T153000
DTSTAMP:20260530T052804
CREATED:20230530T153213Z
LAST-MODIFIED:20230530T153213Z
UID:10000390-1685541600-1685547000@datascience.ucsd.edu
SUMMARY:On the complexity of Frank-Wolfe methods
DESCRIPTION:Abstract: Frank-Wolfe methods are popular for optimization over a polytope. One of the reasons is because they do not need projection onto the polytope but only linear optimization over it. This talk has two parts. \nThe first part will be about the complexity of Wolfe’s method\, an algorithm closely related to Frank-Wolfe methods. In 1974 Phillip Wolfe proposed a method to find the minimum Euclidean-norm point in a convex polyhedron. The method is essentially the same as the Lawson-Hanson algorithm for non-negative least squares. The complexity of Wolfe’s method has remained unknown since he proposed it. The method is important because it is used as a subroutine for one of the most practical algorithms for submodular function minimization. We present the first example that Wolfe’s method takes exponential time. Additionally\, we improve previous results to show that linear programming reduces in strongly-polynomial time to the minimum norm point problem over a simplex. \nThe second part will be about the smoothed complexity of Frank-Wolfe methods. To understand their complexity\, a fruitful approach in many\nworks has been the use of condition measures of polytopes. Lacoste-Julien and Jaggi introduced a condition number for polytopes and showed linear convergence for several variations of the method. The actual running time can still be exponential in the worst case (when the condition number is exponential). We study the smoothed complexity of the condition number\, namely the condition number of small random perturbations of the input polytope and show that it is polynomial for any simplex and exponential for general polytopes. Our argument for polytopes is a refinement of an argument that we develop to study the conditioning of random matrices. The basic argument shows that for c > 1\, a d-by-n random Gaussian matrix with n >= cd has a d-by-d submatrix with minimum singular value that is exponentially small with high probability. This also has consequences on known results about the robust uniqueness of tensor decompositions\, the complexity of the simplex method and the diameter of polytopes.
URL:https://datascience.ucsd.edu/event/on-the-complexity-of-frank-wolfe-methods/
LOCATION:SDSC\, The Auditorium\, 9836 Hopkins Dr\, La Jolla\, San Diego\, CA\, United States
CATEGORIES:Colloquium
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/Los_Angeles:20230530T140000
DTEND;TZID=America/Los_Angeles:20230530T153000
DTSTAMP:20260530T052804
CREATED:20230530T153018Z
LAST-MODIFIED:20230530T153018Z
UID:10000389-1685455200-1685460600@datascience.ucsd.edu
SUMMARY:Representation Learning: A Causal Perspective
DESCRIPTION:Abstract: Representation learning constructs low-dimensional representations to summarize essential features of high-dimensional data like images and texts. Ideally\, such a representation should efficiently capture non-spurious features of the data. It shall also be disentangled so that we can interpret what feature each of its dimensions captures. However\, these desiderata are often intuitively defined and challenging to quantify or enforce. \nIn this talk\, we take on a causal perspective of representation learning. We show how desiderata of representation learning can be formalized using counterfactual notions\, enabling metrics and algorithms that target efficient\, non-spurious\, and disentangled representations of data. We discuss the theoretical underpinnings of the algorithm and illustrate its empirical performance in both supervised and unsupervised representation learning.
URL:https://datascience.ucsd.edu/event/representation-learning-a-causal-perspective/
LOCATION:SDSC\, The Auditorium\, 9836 Hopkins Dr\, La Jolla\, San Diego\, CA\, United States
CATEGORIES:Colloquium
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/Los_Angeles:20230524T140000
DTEND;TZID=America/Los_Angeles:20230524T150000
DTSTAMP:20260530T052804
CREATED:20230323T181207Z
LAST-MODIFIED:20240402T224726Z
UID:10000365-1684936800-1684940400@datascience.ucsd.edu
SUMMARY:Spectral clustering in high-dimensional Gaussian mixture block models
DESCRIPTION:The Gaussian mixture block model is a simple generative model for networks: to generate a sample\, we associate each node with a latent feature vector sampled from a mixture of Gaussians\, and we add an edge between nodes if and only if their feature vectors are sufficiently similar. The different components of the Gaussian mixture represent the fact that there may be several types of nodes with different distributions over features — for example\, in a social network each component represents the different attributes of a distinct community. In this talk I will discuss recent results on the performance of spectral clustering algorithms on networks sampled from high-dimensional Gaussian mixture block models\, where the dimension of the latent feature vectors grows as the size of the network goes to infinity. Our results merely begin to sketch out the information-computation landscape for clustering in these models\, and I will make an effort to emphasize open questions.\nBased on joint work with Shuangping Li.
URL:https://datascience.ucsd.edu/event/tselil-schramm/
LOCATION:SDSC\, The Synthesis Center\, 9500 Gilman Drive\, La Jolla\, CA\, 92093\, United States
CATEGORIES:Colloquium
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/Los_Angeles:20230522T130000
DTEND;TZID=America/Los_Angeles:20230522T143000
DTSTAMP:20260530T052804
CREATED:20230523T143630Z
LAST-MODIFIED:20230523T143643Z
UID:10000388-1684760400-1684765800@datascience.ucsd.edu
SUMMARY:Algorithms for multi-group learning
DESCRIPTION:Abstract: Multi-group agnostic learning is a formal learning criterion that is concerned with the conditional risks of predictors within subgroups of a population. The criterion addresses recent practical concerns such as subgroup fairness and hidden stratification. I’ll talk about the structure of solutions to the multi-group learning problem\, as well as some simple and near-optimal algorithms for the learning problem. This is based on joint work with Christopher Tosh.
URL:https://datascience.ucsd.edu/event/algorithms-for-multi-group-learning/
LOCATION:SDSC\, The Synthesis Center\, 9500 Gilman Drive\, La Jolla\, CA\, 92093\, United States
CATEGORIES:Colloquium
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/Los_Angeles:20230518T140000
DTEND;TZID=America/Los_Angeles:20230518T150000
DTSTAMP:20260530T052804
CREATED:20230323T181111Z
LAST-MODIFIED:20230513T000759Z
UID:10000364-1684418400-1684422000@datascience.ucsd.edu
SUMMARY:Scaling and Generalizing Approximate Bayesian Inference | David Blei
DESCRIPTION:Abstract: A core problem in statistics and machine learning is to approximate difficult-to-compute probability distributions. This problem is especially important in Bayesian statistics\, which frames all inference about unknown quantities as a calculation about a conditional distribution. In this talk I review and discuss innovations in variational inference (VI)\, a method that approximates probability distributions through optimization. VI has been used in myriad applications in machine learning and Bayesian statistics. It tends to be faster than more traditional methods\, such as Markov chain Monte Carlo sampling. \nAfter quickly reviewing the basics\, I will discuss two lines of research in VI. I first describe stochastic variational inference\, an approximate inference algorithm for handling massive datasets\, and demonstrate its application to probabilistic topic models of millions of articles. Then I discuss black box variational inference\, a generic algorithm for approximating the posterior. Black box inference easily applies to many models but requires minimal mathematical work to implement. I will demonstrate black box inference on deep exponential families—a method for Bayesian deep learning—and describe how it enables powerful tools for probabilistic programming. \n  \nBio: David Blei is a Professor of Statistics and Computer Science at Columbia University\, and a member of the Columbia Data Science\nInstitute. He studies probabilistic machine learning\, including its theory\, algorithms\, and application. David has received several awards for his research. He received a Sloan Fellowship (2010)\, Office of Naval Research Young Investigator Award (2011)\, Presidential Early Career Award for Scientists and Engineers (2011)\, Blavatnik Faculty Award (2013)\, ACM-Infosys Foundation Award (2013)\, a Guggenheim fellowship (2017)\, and a Simons Investigator Award (2019). He is the co-editor-in-chief of the Journal of Machine Learning Research. He is a fellow of the ACM and the IMS. \nWebsite : http://www.cs.columbia.edu/~blei/ \n  \nZoom Link : http://bit.ly/HDSI-Seminars
URL:https://datascience.ucsd.edu/event/david-blei/
LOCATION:SDSC\, The Auditorium\, 9836 Hopkins Dr\, La Jolla\, San Diego\, CA\, United States
CATEGORIES:Colloquium,Guest Lecture
ATTACH;FMTTYPE=image/jpeg:https://datascience.ucsd.edu/wp-content/uploads/2023/03/professordavisblei_headshot-scaled.jpg
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/Los_Angeles:20230322T140000
DTEND;TZID=America/Los_Angeles:20230322T153000
DTSTAMP:20260530T052804
CREATED:20230323T182317Z
LAST-MODIFIED:20230323T182317Z
UID:10000370-1679493600-1679499000@datascience.ucsd.edu
SUMMARY:Structured Transformer Models for NLP
DESCRIPTION:Abstract: The field of natural language processing has recently unlocked a wide range of new capabilities through the use of large language models\, such as GPT-4. The growing application of these models motivates developing a more thorough understanding of how and why they work\, as well as further improvements in both quality and efficiency. \nIn this talk\, I will present my work on analyzing and improving the Transformer architecture underlying today’s language models through the study of how information is routed between multiple words in an input. I will show that such models can predict the syntactic structure of text in a variety of languages\, and discuss how syntax can inform our understanding of how the networks operate. I will also present my work on structuring information flow to build radically more efficient models\, including models that can process text of up to one million words\, which enables new possibilities for NLP with book-length text. \nBio: Nikita Kitaev is a final-year Ph.D. student in Computer Science at UC Berkeley\, advised by Dan Klein. His research spans natural language processing and machine learning\, with a focus on understanding the structure and operation of large language models\, including leveraging insights from the field of syntax. He is the recipient of an NSF Graduate Research Fellowship and his work has been recognized through a best paper award at ACL. Previously\, Nikita has received a B.S. in Electrical Engineering and Computer Science from UC Berkeley.
URL:https://datascience.ucsd.edu/event/structured-transformer-models-for-nlp/
LOCATION:SDSC\, The Synthesis Center\, 9500 Gilman Drive\, La Jolla\, CA\, 92093\, United States
CATEGORIES:Colloquium
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/Los_Angeles:20211117T140000
DTEND;TZID=America/Los_Angeles:20211117T150000
DTSTAMP:20260530T052804
CREATED:20211112T223936Z
LAST-MODIFIED:20211112T223936Z
UID:10000338-1637157600-1637161200@datascience.ucsd.edu
SUMMARY:HDSI Seminar Series | Theories of Inference for Visual Analysis
DESCRIPTION:
URL:https://datascience.ucsd.edu/event/hdsi-seminar-series-theories-of-inference-for-visual-analysis/
CATEGORIES:Colloquium,Guest Lecture,HDSI Event,Industry,Seminar,Webinar,Workshops
ATTACH;FMTTYPE=:
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/Los_Angeles:20210408T130000
DTEND;TZID=America/Los_Angeles:20210408T140000
DTSTAMP:20260530T052804
CREATED:20210331T191442Z
LAST-MODIFIED:20210331T191442Z
UID:10000172-1617886800-1617890400@datascience.ucsd.edu
SUMMARY:HDSI Seminar Series: Deep Learning for Market Design: Fairness\, Robustness\, and Expressiveness by John Dickerson
DESCRIPTION:Title\nDeep Learning for Market Design: Fairness\, Robustness\, and Expressiveness \nJohn Dickerson\, Assistant Professor of Computer Science\, University of Maryland; Chief Scientist\, Arthur AI \nAbstractThe design of revenue-maximizing auctions with strong incentive guarantees is a core concern of economic theory. Computational auctions enable online advertising\, sourcing\, spectrum allocation\, and myriad financial markets. Analytic progress in this space is notoriously difficult; since Myerson’s 1981 work characterizing single-item optimal auctions\, there has been limited progress outside of restricted settings. A recent paper by Dütting et al. circumvents analytic difficulties by applying deep learning techniques to\, instead\, approximate optimal auctions. Their RegretNet architecture can represent auctions with arbitrary numbers of items and participants; it is trained to be empirically strategyproof\, but the property is never exactly verified leaving potential loopholes for market participants to exploit. In parallel\, new research from Ilvento et al. and other groups has developed notions of fairness in the context of auction design. Inspired by these advances\, in this talk\, we discuss extensions of these techniques for approximating auctions using deep learning to address concerns of* fairness while maintaining high revenue and strong incentive guarantees;\n* certified robustness\, that is\, verification of claimed strategyproofness of deep learned auctions; and\n* expressiveness via different demand functions and other constraints.To enable that last point\, we propose a new architecture to learn incentive compatible\, revenue-maximizing auctions from sampled valuations\, which uses the Sinkhorn algorithm to perform a differentiable bipartite matching. Our new framework allows the network to learn strategyproof revenue-maximizing mechanisms in settings not learnable by the previous RegretNet architecture. \nThis talk covers hot-off-the-presses work led by PhD students Michael Curry\, Ping-yeh Chiang\, and Samuel Dooley\, and undergraduate students Kevin Kuo\, Uro Lyi\, Anthony Ostuni\, and Elizabeth Horishny. Papers have appeared at NeurIPS-20 or are currently under review; please check arXiv or get in touch for drafts. \nBio\nJohn P Dickerson is an Assistant Professor of Computer Science at the University of Maryland as well as Chief Scientist of Arthur AI\, an enterprise-focused AI/ML model monitoring firm. He is a recipient of awards such as the NSF CAREER Award\, IEEE Intelligent Systems AI’s 10 to Watch\, Google Faculty Research Award\, Google Research Scholar Award\, and paper awards and nominations at venues such as AAAI. His research centers on solving practical economic problems using techniques from computer science\, stochastic optimization\, and machine learning. He has worked extensively on theoretical and empirical approaches to organ exchange where his work has set policy at the UNOS nationwide kidney exchange; worldwide blood donation markets with Facebook; game-theoretic approaches to counter-terrorism and negotiation\, where his models have been deployed; and market design problems in industry (e.g.\, online advertising) through various startups. Dickerson received his PhD in computer science from Carnegie Mellon University.
URL:https://youtu.be/nWmI2bs_mcs#new_tab
CATEGORIES:Colloquium,HDSI Event,Seminar
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/Los_Angeles:20210401T130000
DTEND;TZID=America/Los_Angeles:20210401T140000
DTSTAMP:20260530T052804
CREATED:20210331T183837Z
LAST-MODIFIED:20210331T183837Z
UID:10000171-1617282000-1617285600@datascience.ucsd.edu
SUMMARY:HDSI Seminar Series: On the Implicit Bias of Stochastic Gradient Descent with Moderate Learning Rate by Quanquan Gu
DESCRIPTION:Title: On the Implicit Bias of Stochastic Gradient Descent with Moderate Learning Rate \nHDSI Seminar Series \nQuanquan Gu\, Assistant Professor of Computer Science at UCLA \nAbstract: Understanding the algorithmic bias of stochastic gradient descent (SGD) is one of the key challenges in modern machine learning and deep learning theory. Most of the existing works\, however\, focus on very small or even infinitesimal learning rate regimes and fail to cover practical scenarios where the learning rate is moderate and annealing. In this talk\, I will introduce our attempt to characterize the particular regularization effect of SGD in the moderate learning rate regime by studying its behavior for optimizing an overparameterized linear regression problem. In this case\, SGD and GD are known to converge to the unique minimum-norm solution; however\, with the moderate and annealing learning rate\, we show that they exhibit different directional bias: SGD converges along the large eigenvalue directions of the data matrix\, while GD goes after the small eigenvalue directions. Furthermore\, we show that such directional bias does matter when early stopping is adopted\, where the SGD output is nearly optimal but the GD output is suboptimal. Our analysis explains several folk arts in practice used for SGD hyperparameter tuning\, such as (1) linearly scaling the initial learning rate with batch size; and (2) overrunning SGD with a high learning rate even when the loss stops decreasing. \nThis talk is based on joint work with Jingfeng Wu\, Difan Zou\, and Vladimir Braverman. \nBio: Quanquan Gu is an Assistant Professor of Computer Science at UCLA. His current research is in the area of artificial intelligence and machine learning\, with a focus on developing and analyzing nonconvex optimization algorithms for machine learning and building the theoretical foundations of deep learning. He received his Ph.D. degree in Computer Science from the University of Illinois at Urbana-Champaign in 2014. He is a recipient of the Yahoo! Academic Career Enhancement Award\, NSF CAREER Award\, Simons Berkeley Research Fellowship\, Adobe Data Science Research Award\, Salesforce Deep Learning Research Award\, and AWS Machine Learning Research Award.
URL:https://youtu.be/qLGvn4RJwl4#new_tab
CATEGORIES:Colloquium,HDSI Event,Seminar
ATTACH;FMTTYPE=:
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/Los_Angeles:20201106T080000
DTEND;TZID=America/Los_Angeles:20201106T170000
DTSTAMP:20260530T052804
CREATED:20201105T192203Z
LAST-MODIFIED:20201105T192203Z
UID:10000283-1604649600-1604682000@datascience.ucsd.edu
SUMMARY:Deep-Math Conference 2020
DESCRIPTION:
URL:https://datascience.ucsd.edu/event/deep-math-conference-2020/2020-11-06/
CATEGORIES:Colloquium,Conference,HDSI Event,Webinar,Workshops
ATTACH;FMTTYPE=:
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/Los_Angeles:20201105T080000
DTEND;TZID=America/Los_Angeles:20201105T170000
DTSTAMP:20260530T052804
CREATED:20201105T192203Z
LAST-MODIFIED:20201105T192203Z
UID:10000282-1604563200-1604595600@datascience.ucsd.edu
SUMMARY:Deep-Math Conference 2020
DESCRIPTION:
URL:https://datascience.ucsd.edu/event/deep-math-conference-2020/2020-11-05/
CATEGORIES:Colloquium,Conference,HDSI Event,Webinar,Workshops
ATTACH;FMTTYPE=:
END:VEVENT
BEGIN:VEVENT
DTSTART;TZID=America/Los_Angeles:20200211T140000
DTEND;TZID=America/Los_Angeles:20200211T150000
DTSTAMP:20260530T052804
CREATED:20200123T223404Z
LAST-MODIFIED:20200123T223404Z
UID:10000052-1581429600-1581433200@datascience.ucsd.edu
SUMMARY:Tal Linzen
DESCRIPTION:
URL:https://datascience.ucsd.edu/event/tal-linzen/
LOCATION:3234 Matthews Ln\, La Jolla\, 92093\, United States
CATEGORIES:Colloquium
END:VEVENT
END:VCALENDAR