EVENTS

EVENTS

EnCORE is hosting a 3-day workshop to reinvigorate collaboration between the approximation and computational geometry communities. To learn more about the workshop, visit here

To attend this workshop virtually, register via Eventbrite here. Details will be provided upon registration.

 

Alex Dimakis, a professor at UT Austin and the co-director of the National AI Institute on the Foundations of Machine Learning, will be presenting the talk “Deep Generative Models and Inverse Problems for Signal Reconstruction” on April 1, 2024, as a part of the Foundations of Data Science – Virtual Talk Series.

Register to attend this talk here! The Zoom link will be sent upon registration. 

Join the 2-day EnCORE tutorial “Characterizing and Classifying Cell Types of the Brain: An Introduction for Computational Scientists” with Michael Hawrylycz, Ph.D., Investigator, Allen Institute. 

This tutorial will take place on March 21 – 22, 2024 at UC San Diego in the Computer Science and Engineering Building in Room 1242. There will be a Zoom option available as well. 

Register for this tutorial here

Learn more about the tutorial by visiting this event page. 

 

Join EnCORE’s talk on “Theoretical Exploration of Foundation Model Adaptation Methods” with Kangwook Lee, an assistant professor in the Electrical and Computer Engineering Department and the Computer Sciences Department at the University of Wisconsin-Madison.

This talk will take place on Friday, February 9, 2024, at UC San Diego, on the 4th floor of Atkinson Hall. 

Alexander “Sasha” Rush is an Associate Professor at Cornell Tech, where he studies natural language processing and machine learning. Sasha received his PhD from MIT supervised by Michael Collins and was a postdoc at Facebook NY under Yann LeCun. His research interest is in deep learning text generation, generative modeling, and structured prediction. His group also supports open-source development, including the OpenNMT machine translation system. His research has received several best paper awards at NLP conferences and an NSF Career award. 

Monday, Feb 5th – 1pm PT (4pm ET, 9pm UTC) 
 
Sasha Rush (Cornell) 

 

Learn more about this upcoming talk here.

The NSF TRIPODS Workshop brings together a large and diverse team of researchers from each of the TRIPODS centers: EnCOREFODSIIDEAL, and IFDS. The aim of the Transdisciplinary Research In Principles Of Data Science (TRIPODS) is to connect the statistics, mathematics, and theoretical computer science communities to develop the theoretical foundations of data science through integrated research and training activities focused on core algorithmic, mathematical, and statistical principles. This workshop focuses on the contributions of each center’s work and will take place February 1 – 2, 2024. 

FinDS is a 3-week summer program that is designed to provide high school students a strong base in the Mathematical Foundations of Data Science, with in-person and virtual options available. 

FinDS offers two tracks: Beginners and Advanced. Applications for the two tracks are live now.  Learn more about these two tracks below and visit mathfinds.ucsd.edu for more details. 

The application for the Beginner’s Program can be found here

The application for the Advanced Program can be found here

Apply for the program online by May 1, 2024.

Join EnCORE’s talk on “Tight Bound on Equivalence Testing in Conditional Sampling Model” with Diptarka Chakraborty, an assistant professor at the National University of Singapore.

This talk will take place on Monday, December 4, 2023 at UC San Diego in the CSE Building in Room 4258 and Zoom. 

Join EnCORE’s talk on “Bounded Weighted Edit Distance” with Tomasz Kociumaka, a postdoctoral researcher at the Max Planck Institute for Informatics, Germany.

This talk will take place on Monday, November 20, 2023 at UC San Diego in the CSE Building in Room 4258 and Zoom. 

Join EnCORE’s talk on “Fredman’s Trick Meets Dominance Product: Fine-Grained Complexity of Unweighted APSP, 3SUM Counting, and More” with Yinzhan Xu, a PhD student studying theoretical computer science at MIT. 

This talk will take place on Monday, Monday, October 23, 2023, at UC San Diego in the CSE Building in Room 4258.

Boaz Barak is the Gordon McKay professor of Computer Science at Harvard University’s John A. Paulson School of Engineering and Applied Sciences. Barak’s research interests include all areas of theoretical computer science and in particular cryptography, computational complexity, and the foundations of machine learning. Previously, he was a principal researcher at Microsoft Research New England, and before that an associate professor (with tenure) at Princeton University’s computer science department. Barak has won the ACM dissertation award, the Packard and Sloan fellowships, and was also selected for Foreign Policy magazine’s list of 100 leading global thinkers for 2014.

Monday, Oct 30 – 1pm PT (4pm ET, 9pm UTC) 
 
Boaz Barak (Harvard)

 

Learn more about this upcoming talk here.

 

EnCORE students have teamed up to collaborate on various topics, such as augmented algorithms and large language models! 

This workshop will take place on Wednesday, October 18 at 9 a.m. (PST) via Zoom. 

EnCORE and the Data Science Institute (DSI) at the University of Chicago are partnering to celebrate the potential of exceptional data scientists through an intensive research and career workshop over the course of two days from Monday, November 13, 2023 to Tuesday, November 14, 2023.  This workshop will take place at the University of Chicago. 


Apply by September 22nd at 11:59pm CDT to be considered. 

Join us and listen to the NSF TRIPODS Panel discussion on the reliability of Large Language Models (LLMs).

The panelists for this session include Marzyeh Ghassemi (MIT), James Zou (Stanford), Ernest Davis (NYU), and Nisheeth Vishnoi (Yale).

Tuesday, Sept 5, 2023 – 2pm PT (5pm ET, 10pm UTC) 

 

 
 

EnCORE is organizing a workshop on the statistical and computational requirements for solving various learning problems from Monday, February 26, 2024 to Friday, March 1, 2024. This event aims to provide a forum to discuss the latest research and develop new ideas, as well as build bridges between different disciplines, such as applied mathematics, statistics, optimization, and theoretical computer science. 

 

EnCORE is hosting a two-day conference from Thursday, June 1, 2023, to Friday, June 2, 2023. This event aims to unite scientists from diverse fields, providing a platform to engage with a broad range of students. The conference will be held at UC San Diego’s CSE Building in Room 1242. 

Don’t miss out on this unique opportunity – secure your spot and choose your preferred tickets for each day by registering here.

TCS for All is holding its TCS for All Spotlight Workshop on Thursday, June 22nd, 2023 (2-4pm), in Orlando, Florida, USA, as part of the 54th Symposium on Theory of Computing (STOC) and TheoryFest. The workshop is open to all.

To attend the workshop in person, just show up! Otherwise, to attend remotely, registration is free, but you must register to get the zoom link. 

Register for this workshop here.

Join EnCORE’s 2-day tutorial on Statistical Machine Learning Models for Neural and Behavioral Data Analysis with Anqi Wu, an assistant professor in the Computer Science and Engineering (CSE) department at the Georgia Institute of Technology. This tutorial is geared toward Computer Science and Data Science junior graduate students with little to no experience in neuroscience.

This tutorial will take place on Monday, May 22, 2023 to Tuesday, May 23, 2023 at UC San Diego in the CSE Building in Room 1242. 

Register for this tutorial here

Learn more about the tutorial by visiting the Neuroscience Tutorial page of Events.  

This workshop will address research into topics in computational complexity, cryptographic assumptions, randomness in computation, machine learning, complexity of proofs, and related topics that address issues raised by the five worlds conundrum.

This workshop will take place on Friday, May 12, 2023 to Saturday, May 13, 2023 at UC San Diego in the Computer Science and Engineering (CSE) Building. Visit wcw.ucsd.edu for more details about the event.

FinDS is an innovative program that is designed to provide high school students with a strong base in the Mathematical Foundations of Data Science over the course of 3 weeks. This program is offered both virtually and in-person. 

Learn more about the program at mathfinds.ucsd.edu or visiting the FinDS page of Outreach.

Apply by June 1, 2023 to be considered.

 

EnCORE Institute brings together students to meet biweekly to discuss research topics in an informal setting. Join us to expand your knowledge on various subjects while meeting like-minded individuals who are also part of EnCORE.

 

Vladimir Braverman is a Professor of Computer Science at Rice University and a Visiting Researcher with Google Research. Previously, he was an Associate Professor in the Department of Computer Science at Johns Hopkins University.  Braverman received a PhD in Computer Science at UCLA. Braverman’s research interests include efficient sublinear algorithms and their applications to machine learning and digital health.

Thursday, March 16, 2023 – 1pm PT (4pm ET, 9pm UTC)

Vladimir Braverman (Rice University)
 

EnCORE Institute was invited to organize a workshop on the theme of Distributed Learning and Decision Making across four sessions at the 2023 Information Theory and Applications workshop (ITA), a casual gathering for researchers that apply theory to diverse areas in science and engineering. ITA was held from Sunday, February 12, 2023 to Friday, February 17, 2023.

 

Amit Chakrabarti is a professor at Dartmouth College in the Department of Computer Science. His research focuses on theoretical computer science. Specifically, this includes complexity theory, data stream algorithms, and approximation algorithms. He is a recipient of the NSF CAREER Award, Karen E. Wetterhahn Award, the McLane Family Fellowship, and the Friedman Family Fellowship.

Thursday, Feb 9, 2023 – 2pm PT (5pm ET, 10pm UTC)

Amit Chakrabarti (Dartmouth)

EnCORE brings together scientists from multiple disciplines to reach a wide demography of students in this one-day mini-conference.

Monday, December 19, 2022

Register by December 15, 2022

Tickets are FREE for EnCORE Affiliated Faculty, Program Participants, all CSE and HDSI faculty and students and also for all remote participants, and $55 for Guests who are attending in person.

Omri Ben-Eliezer is an Applied Mathematics instructor at MIT who is affiliated with the Foundations of Data Science Institute. He conducted research under the guidance of Noga Alon at Tel Aviv University, Moni Naoor at Weizmann Institute, and Mandhu Sudan at Harvard University. His research interests center around the theoretical and algorithmic foundations of big data, which include sublinear time and streaming algorithms, robustness and privacy, knowledge representation, and complex networks.

Wednesday Dec 14, 2022 – 1pm PT (4pm ET, 10pm CEST, 8pm UTC) 

Omri Ben-Eliezer (MIT)

Edith Cohen is a research scientist at Google and a professor at Tel-Aviv University in the School of Computer Science. Before that, she served as a member of the research staff at AT&T Labs Research and the principal researcher at Microsoft Research. Her research interests include algorithms design, data mining, machine learning, optimization, and networks. In her research, she has developed models and scalable algorithms in a variety of areas, including query processing and optimization, caching, routing, and streaming. Cohen received the IEEE Communications Society William R. Bennett Prize in 2007 and was recognized as an ACM Fellow in 2017.

Saturday, November 12, 2022 – 1pm PT (4pm ET, 10pm CEST, 8pm UTC)
Edith Cohen (Google Research / Tel-Aviv University)

David Woodruff is a professor at Carnegie Mellon University in the Computer Science Department. Before that he was a research scientist at the IBM Almaden Research Center. His research interests include data stream algorithms, distributed algorithms, machine learning, numerical linear algebra, optimization, sketching, and sparse recovery. He is the recipient of the 2020 Simons Investigator Award, the 2014 Presburger Award, and Best Paper Awards at STOC 2013, PODS 2010, and PODS, 2020. At IBM he was a member of the Academy of Technology and a Master Inventor.

 

Wednesday Oct 5 – 1pm PT (4pm ET, 10pm CEST, 8pm UTC)
David Woodruff (CMU)

For the first time since 2019 #DeepMath will be in-person, in November in San Diego 😎 Abstract submissions are open deadline August 15. We are looking forward for great submissions focusing on theory of deep networks https://deepmath-conference.com/submissions
@deepmath1

A virtual series that brings prominent leaders in Foundations of Data Science across Computer Science, Engineering, Mathematics, and Statistics.

Data Science Fall retreat will take place on December 10th, 2021 and will bring together researchers from four NSF TRIPODS Institutes, students, and postdocs. Detailed program here.

Many modern applications produce massive datasets containing a lot of redundancy, either in the form of highly skewed frequencies or repeating motifs/fragments of identical data. Recent years have seen a lot of activities on new data compression techniques, and computation over compressed data. However, there has been little cross-community interaction between theoreticians and those working with real data. Computation+Compression workshop, 2022 aims to bridge the gap by bringing together the experts from Theoretical Computer Science, and Bioinformatics community that work in this area.

In recent years, the computational paradigm for large-scale machine learning and data analytics has shifted to a massively large distributed system composed of individual computational nodes (e.g., ranges from GPUs to low-end commodity hardware). A typical distributed optimization problem for training machine learning models has several considerations. The first of these is the convergence rate of the algorithm. While the convergence rate is a measure of efficiency in conventional systems, in today’s distributed computing framework, the aspects of communication are at odds with the convergence rate.  In recent years various techniques have been developed to alleviate the problem of communication bottleneck, including but not limited to quantization and sparsification, local updates, one-shot learning. The goal of this workshop is to bring these ideas together.

Past Events:

Shuchi Chawla on Foundation of Data Science Series, April 6, 2022

Eric Tchetgen Tchetgen on Foundation of Data Science Series, March 16, 2022

Computation+Compression workshop to be held on Jan 19th, 2022. Register to participate.

Piotr Indyk, MIT on Foundations of Data Science Virtual Series: Learning Based Sampling and Streaming, Jan 13th, 2022.

Data Science Annual Retreat, Dec 10th, 2021

Undergraduate Capstone Project Presentation, Dec , 2021

Nicole Immorlica, Microsoft Research, on Foundations of Data Science Virtual Series: Communicating with Anecdotes, Nov 16th, 2021 

Maxim Raginsky, UIUC, on Foundations of Data Science Virtual Series: Neural SDEs: Deep Generative Models in the Diffusion Limits, Oct 21st, 2021 

Christos Papadimitrou, Columbia University, on Foundations of Data Science Virtual Series: How Does the Brain begets the Mind?, Sept 23rd, 2021

Hamed Hassani, University of Pennsylvania on  Foundations of Data Science Virtual Series: Learning Robust Models, May 6th, 2021

Workshop on Communication Efficient Distributed Optimization, April 9th, 2021

Product Manifold Learning with Independent Coordinate Selection

Jesse He (UCSD)

Abstract: In many dimensionality reduction tasks, we wish to identify the constituent components that explain our observations. For manifold learning, this can be formalized as factoring a Riemannian product manifold. Recovering this factorization, however, may suffer from certain difficulties in practice, especially when data is sparse or noisy, or when one factor is distorted by the other. To address these limitations, we propose identifying non-redundant coordinates on the product manifold before applying product manifold learning to identify which coordinates correspond to different factor manifolds.

Tear and Repulsion Enabled Registration of Point Clouds for Manifold Learning

Dhruv Kohli (UCSD)

Abstract: We present a framework for aligning the local views of a possibly closed/non-orientable data manifold to produce an embedding in its intrinsic dimension through tearing. Through a spectral coloring scheme, we render the embeddings of the points across the tear with matching colors, enabling a visual recovery of the topology of the data manifold. The embedding is further equipped with a tear-aware metric that enables computation of shortest paths while accounting for the tear. To measure the quality of an embedding, we propose two Lipschitz-type notions of global distortion—a stronger and a weaker one—along with their pointwise counterparts for a finer assessment of the embedding. Subsequently, we bound them using the distortion of the local views and the alignment error between them. We show that our theoretical result on strong distortion leads to a new perspective on the need for a repulsion term in manifold learning objectives. As a result, we enhance our alignment approach by incorporating repulsion. Finally, we compare various strategies for the tear and repulsion enabled alignment, with regard to their speed of convergence and the quality of the embeddings produced.


This is joint work with my advisors Gal Mishne and Alex Cloninger at UCSD.

Efficient Online Clustering with Moving Costs

Dimitrios Christou (UT Austin)

Abstract: In this talk, I will consider an online-learning problem, called Online Clustering with Moving Costs, at which a learner maintains a set of facilities over rounds so as to minimize the connection cost of an adversarially selected sequence of clients. The learner is informed on the positions of the clients at each round only after its facility-selection and can use this information to update its decision in the next round. However, updating the facility positions comes with an additional moving cost based on the moving distance of the facilities. I will be presenting the first (polynomial-time) approximate-regret algorithm for this setting through a combination of different algorithmic techniques such as HST embeddings, the FTRL framework with a dilated entropic regulariser as well as a novel rounding scheme.

Encoding Structural Symmetry is Key for Length Generalization in Arithmetic Tasks

Mahdi Sabbaghi (UPenn)

Abstract: Despite the success of Transformers on language understanding, code generation, and logical reasoning, they still fail to (length) generalize on basic arithmetic tasks such as addition and multiplication. A major reason behind this failure is the vast difference in structure of numbers versus text; for example, numbers are associated with a specific significance order that plays a role in calculating the answer. In contrast, for text, such symmetries are quite unnatural. In this work, we propose to encode these semantics explicitly into the model via appropriate data formatting and custom positional encodings. To further elucidate the importance of explicitly encoding structure, in a simplified linear setting, we prove that standard positional encodings even when trained with augmentations to implicitly induce structure fail at such generalization, whereas enforcing structure via positional encodings succeeds.

Bio: Mahdi Sabbaghi is a second year PhD student at UPenn, department of Electrical and System Engineering, supervised by Professors Hamed Hassani and George Pappas. Previously he obtained a B. Sc. degree in Electrical Engineering as well as a B. Sc. degree in Physics from the Sharif University of Technology, in Tehran.

Private Estimation of U Statistics

Shourya Pandey (UT Austin)

Abstract: We consider the problem of private estimation of U statistics. U statistics are widely used estimators that naturally arise in a broad class of problems, from nonparametric signed rank tests to subgraph counts in random networks. Despite the recent outpouring of interest in private mean estimation, private algorithms for more general U statistics have received little attention.  We propose a framework where, for a broad class of U statistics, one can use existing tools in private mean estimation to obtain confidence intervals where the private error does not overwhelm the irreducible error resulting from the variance of the U statistics. However, in specific cases that arise when the U statistics degenerate or have vanishing moments, the private error may be of a larger order than the non-private error. To remedy this, we propose a new thresholding-based approach that uses Hajek projections to re-weight different subsets. As we show,  this leads to more accurate inference in certain settings.

Finite-Time Logarithmic Bayes Regret Upper Bounds

Alexia Atsidakou (UT Austin)

Abstract: We derive the first finite-time logarithmic Bayes regret upper bounds for Bayesian bandits, for BayesUCB and Thompson Sampling. In Gaussian and Bernoulli multi-armed bandits, we obtain $O(c_\Delta \log n)$ and $O(c_h \log^2 n)$ upper bounds for an upper confidence bound algorithm, where $c_h$ and $c_\Delta$ are constants depending on the prior distribution and the gaps of bandit instances sampled from it, respectively. The latter bound asymptotically matches the lower bound of Lai (1987). Our proofs are a major technical departure from prior works, while being simple and general. The key idea in our proofs is to conduct a frequentist per-instance analysis with Bayesian confidence intervals, and then integrate it over the prior.

Our results provide insights on the value of prior in the Bayesian setting, both in the objective and as a side information given to the learner. They significantly improve upon existing $\tilde{O}(\sqrt{n})$ bounds, which have become standard in the literature despite the logarithmic lower bound of Lai (1987).

Pareto-Optimal Algorithms for Learning in Games

Natalie Collina and Eshwar Ram Arunachaleswaran

Abstract: We study the problem of characterizing optimal learning algorithms for playing repeated games against an adversary with unknown payoffs. In this problem, the first player (called the learner) commits to a learning algorithm against a second player (called the optimizer), and the optimizer best-responds by choosing the optimal dynamic strategy for their (unknown but well-defined) payoff. Classic learning algorithms (such as no-regret algorithms) provide some counterfactual guarantees for the learner, but might perform much more poorly than other learning algorithms against particular optimizer payoffs.

In this paper, we introduce the notion of asymptotically Pareto-optimal learning algorithms. Intuitively, if a learning algorithm is Pareto-optimal, then there is no other algorithm which performs asymptotically at least as well against all optimizers and performs strictly better (by at least $\Omega(T)$) against some optimizer. We show that well-known no-regret algorithms such as Multiplicative Weights and Follow The Regularized Leader are Pareto-dominated. However, while no-regret is not enough to ensure Pareto-optimality, we show that a strictly stronger property, no-swap-regret, is a sufficient condition for Pareto-optimality.

Proving these results requires us to address various technical challenges specific to repeated play, including the fact that there is no simple characterization of how optimizers who are rational in the long-term best-respond against a learning algorithm over multiple rounds of play. To address this, we introduce the idea of the asymptotic menu of a learning algorithm: the convex closure of all correlated distributions over strategy profiles that are asymptotically implementable by an adversary. Interestingly, we show that all no-swap-regret algorithms share the same asymptotic menu, implying that all no-swap-regret algorithms are “strategically equivalent”.

This talk is based on work with Jon Schneider.

Metric Learning from Lazy Crowds

Geelon So

Abstract: We consider crowd-based metric learning from preference comparisons, where given two items, a user prefers the item that is closer to their latent ideal item. Here, “closeness” is measured with respect to a shared but unknown Mahalanobis distance. Can we recover this distance when we can only obtain very few responses per user?

In this very low-budget regime, we show that generally, nothing at all about the metric is revealed, even with infinitely many users. But when the items have subspace cluster structure, we present a divide-and-conquer approach for metric recovery, and provide theoretical recovery guarantees and empirical validation.

This is joint work with Zhi Wang (UCSD) and Ramya Korlakai Vinayak (UW–Madison).

Random Walks, Conductance, and Resistance for the Connection Graph Laplacian

Zhengchao Wan

Abstract: We investigate the concept of effective resistance in connection graphs, expanding its traditional application from undirected graphs. We propose a robust definition of effective resistance in connection graphs by focusing on the duality of Dirichlet-type and Poisson-type problems on connection graphs. Additionally, we delve into random walks, taking into account both node transitions and vector rotations. This approach introduces novel concepts of effective conductance and resistance matrices for connection graphs, capturing mean rotation matrices corresponding to random walk transitions. Thereby, it provides new theoretical insights for network analysis and optimization.

This is based on a joint work with Alexander Cloninger, Gal Mishne, Andreas Oslandsbotn, Sawyer Jack Robertson and Yusu Wang.

Approximability and Inapproximability of Strict-CSPs

Akbar Rafiey

Abstract: We study the approximability and inapproximability of Strict-CSPs. An instance of the Strict-CSPs consists of a set of constraints over a set of variables and a cost function over the assignments. The goal is to find an assignment to the variables of minimum cost which satisfies all the constraints. Some prominent problems that this framework captures are (Hypergraph) Vertex Cover, Min Sum k-Coloring, Multiway Cut, Min Ones, and others.

We focus on a systematic study of Strict-CSPs of the form Strict-CSPs(H), that is, Strict-CSPs where the type of constraints is limited to predicates from a set H. Our first result is a dichotomy for approximation of Strict-CSPs(H), where H is a binary predicate, i.e., a digraph. We prove that if digraph H has bounded width, then Strict-CSPs(H) is approximable within a constant factor (depending on H); otherwise, there is no approximation for Strict-CSPs(H) unless P=NP.

Second, we study the inapproximability of Strict-CSP and present the first general hardness of approximation for Strict-CSP. More precisely, we prove a dichotomy theorem that states every instance of Strict-CSP(H) (H being a digraph) is either polynomial-time solvable or APX-complete. Moreover, we show the existence of a universal constant 0<\delta<1 such that it is NP-hard to approximate Strict-CSP(H) within a factor of (2-\delta) for all digraphs H where Strict-CSP(H) is NP-complete.

Buy-many Mechanisms for Many Unit-demand Buyers

Rojin Rezvan

Abstract: A recent line of research has established a novel desideratum for designing approximatelyrevenue-optimal multi-item mechanisms, namely the buy-many constraint. Under this constraint, prices for different allocations made by the mechanism must be subadditive implying that the price of a bundle cannot exceed the sum of prices of individual items it contains. This natural constraint has enabled several positive results in multi-item mechanism design bypassing well-established impossibility results. Our work addresses a main open question from this literature involving the design of buymany mechanisms for multiple buyers. Our main result is that a simple sequential item pricing mechanism with buyer-specific prices can achieve an O(log m) approximation to the revenue of any buy-many mechanism when all buyers have unit-demand preferences over m items. This is the best possible as it directly matches the previous results for the single-buyer setting where no simple mechanism can obtain a better approximation. Our result applies in full generality: even though there are many alternative ways buy-many mechanisms can be defined for multibuyer settings, our result captures all of them at the same time. We achieve this by directly competing with a more permissive upper-bound on the buy-many revenue, obtained via an ex-ante relaxation.

Streaming PCA for Markovian Data

Syamantak Kumar

Abstract: Since its inception in 1982, Oja’s algorithm has become an established method for streaming principle component analysis (PCA). We study the problem of streaming PCA, where the data-points are sampled from an irreducible, aperiodic, and reversible Markov chain. Our goal is to estimate the top eigenvector of the unknown covariance matrix of the stationary distribution. This setting has implications in scenarios where data can solely be sampled from a Markov Chain Monte Carlo (MCMC) type algorithm, and the objective is to perform inference on parameters of the stationary distribution. Most convergence guarantees for Oja’s algorithm in the literature assume that the data-points are sampled IID. For data streams with Markovian dependence, one typically downsamples the data to get a “nearly” independent data stream. In this paper, we obtain the first sharp rate for Oja’s algorithm on the entire data, where we remove the logarithmic dependence on the sample size, resulting from throwing data away in downsampling strategies.

A d^{1/2 + o(1)} Monotonicity Tester for Boolean Functions on d-Dimensional Hypergrids

Hadley Black

Abstract: Monotonicity testing of Boolean functions over the n-width, d-dimensional hypergrid is a classic problem in property testing, where the goal is to design a randomized algorithm which can distinguish monotone functions from those which are far from any monotone function while making as few queries as possible. The special case of n = 2 corresponds to the hypercube domain. Here a long line of works exploiting a very interesting connection with isoperimetric inequalities for Boolean functions culminated in a non-adaptive tester making ~O(d^{1/2}) queries in a celebrated paper by Khot, Minzer, and Safra (SICOMP 2018). This is known to be optimal for non-adaptive testers. However, the general case of hypergrids for n > 2 remained open. Very recently, two papers (Black-Chakrabarty-Seshadhri STOC 2023 and Braverman-Khot-Kindler-Minzer ITCS 2023) independently obtained ~O(poly(n) d^{1/2}) query testers for hypergrids. These results are essentially optimal for n < polylog(d), but are far from optimal for n >> polylog(d).

This talk covers our most recent result (appearing at FOCS 2023) which obtains a non-adaptive d^{1/2+o(1)} query tester for all n, resolving the non-adaptive monotonicity testing problem for hypergrids, up to a factor of d^{o(1)}. Our proof relies on many new techniques as well as two key theorems which we proved in earlier works from SODA 2020 and STOC 2023.

SmoothLLMs: Defending LLMs against Adversarial Attacks

Alex Robey

Abstract: Despite efforts to align large language models (LLMs) with human values, widely-used LLMs such as GPT, Llama, Claude, and PaLM are susceptible to jailbreaking attacks, wherein an adversary fools a targeted LLM into generating objectionable content.  To address this vulnerability, we propose SmoothLLM, the first algorithm designed to mitigate jailbreaking attacks on LLMs.  Based on our finding that adversarially-generated prompts are brittle to character-level changes, our defense first randomly perturbs multiple copies of a given input prompt, and then aggregates the corresponding predictions to detect adversarial inputs.  SmoothLLM reduces the attack success rate on numerous popular LLMs to below one percentage point, avoids unnecessary conservatism, and admits provable guarantees on attack mitigation.  Moreover, our defense uses exponentially fewer queries than existing attacks and is compatible with any LLM.

Composition of Nested Embeddings with an Application to Outlier Removal

Kristin Sheridan

Abstract: We study the design of embeddings into Euclidean space with outliers. Given a metric space $(X,d)$ and an integer $k$, the goal is to embed all but $k$ points in $X$ (called the “”outliers””) into $\ell_2$ with the smallest possible distortion $c$. Finding the optimal distortion $c$ for a given outlier set size $k$, or alternately the smallest $k$ for a given target distortion $c$ are both NP-hard problems. In fact, it is UGC-hard to approximate $k$ to within a factor smaller than $2$ even when the metric sans outliers is isometrically embeddable into $\ell_2$. We consider bi-criteria approximations. Our main result is a polynomial time algorithm that approximates the outlier set size to within an $O(\log^2 k)$ factor and the distortion to within a constant factor.

The main technical component in our result is an approach for constructing a composition of two given embeddings from subsets of $X$ into $\ell_2$ which inherits the distortions of each to within small multiplicative factors. Specifically, given a low $c_S$ distortion embedding from $S\subset X$ into $\ell_2$ and a high(er) $c_X$ distortion embedding from the entire set $X$ into $\ell_2$, we construct a single embedding that achieves the same  distortion $c_S$ over pairs of points in $S$ and an expansion of at most $O(\log k)\cdot c_X$ over the remaining pairs of points, where $k=|X\setminus S|$. Our composition theorem extends to embeddings into arbitrary $\ell_p$ metrics for $p\ge 1$, and may be of independent interest. While unions of embeddings over disjoint sets have been studied previously, to our knowledge, this is the first work to consider compositions of {\em nested} embeddings.

Graph Sparsification by Approximate Matrix Multiplication

Neo Charalambides

Abstract:  Graphs arising in statistical problems, signal processing, large networks, combinatorial optimization, and data analysis are often dense, which causes both computational and storage bottlenecks. One way of sparsifying a weighted graph, while sharing the same vertices as the original graph but reducing the number of edges, is through spectral sparsification. We study this problem through the perspective of RandNLA. Specifically, we utilize randomized matrix multiplication to give a clean and simple analysis of how sampling according to edge weights gives a spectral approximation to graph Laplacians, without requiring spectral information. Through the CR−MM algorithm, we attain a simple and computationally efficient sparsifier whose resulting Laplacian estimate is unbiased and of minimum variance. Furthermore, we define a new notion of additive spectral sparsifiers, which has not been considered in the literature.

Talk Title

Student Name

Abstract