- talks - lists.cs.princeton.edu

Next Week's Virtual Wilks Statistics Seminar: Monday, March 8th at 12:30pm EST: Daniel Soudry, Technion
by Emily C. Lawrence 02 Mar '21

02 Mar '21

***Next Week’s Virtual Wilks Statistics Seminar*** DATE: Monday, March 8th, 2021 TIME: 12:30pm EST LOCATION: Virtual LINK: [ https://princeton.zoom.us/j/91215738939?pwd=RzNISGMxMTMydG5ZdjFEcS9aRnF4dz09 | https://princeton.zoom.us/j/91215738939?pwd=RzNISGMxMTMydG5ZdjFEcS9aRnF4dz09 ] SPEAKER: Daniel Soudry, Technion TITLE : Surprises in Deep Learning Training ABSTRACT: We make a few interesting observations regarding Deep Neural Networks (DNNS) training: 1. We find empirically that the neural gradients (i.e., backpropagated errors) have a near-lognormal distribution during training. We show how this can be used to analytically derive pruning thresholds and quantization formats - which enable efficient training. 2. DNNs are typically initialized with (a) random weights in order to (b) break symmetry and (c) promote feature diversity. We demonstrate a, b, and c are not necessary at initialization, in order to obtain high accuracy at the end of the training. 3. Quantization of full precision trained DNNs while retaining high accuracy typically requires "fine-tuning" the model on a large dataset. We generate synthetic data from the DNN Batch-Norm statistics; we then use it "fine-tune" the DNN to good performance, thus eliminating the need for real data . 4. Asynchronous training causes a degradation in generalization, even after training has converged to a steady-state. We close this gap by adjusting hyper-parameters according to a theoretical framework aiming to maintain minima stability. Bio: Daniel is an assistant professor and in the Electrical Engineering Department at the Technion, working in the areas of machine learning and neural networks. His recent works focus on resource efficiency and implicit bias in neural networks. He did his post-doc working with Prof. Liam Paninski in the Department of Statistics and the Center for Theoretical Neuroscience at Columbia University, and his Ph.D. in the Electrical Engineering Department at the Technion. He is the recipient of the Gruss Lipper Fellowship, the Taub Fellowship, the Goldberg Award, and Intel's Rising Star Faculty Award.

1 0

3/4/2021 ECE Seminar - Xiaojing Liao, Indiana University Bloomington
by Emily C. Lawrence 02 Mar '21

02 Mar '21

ECE SEMINAR Speaker: Xiaojing Liao, Indiana University Bloomington Title: Toward Automatically Evaluating Security Risks and Providing Cyber Threat Intelligence Day: Thursday, March 4, 2021 Time: 4:30 pm Room: Please register at [ https://princeton.zoom.us/meeting/register/tJEuduyrpjkqGtbe3yU5w-iYC0HxIIcM… | https://princeton.zoom.us/meeting/register/tJEuduyrpjkqGtbe3yU5w-iYC0HxIIcM… ] Host: Prateek Mittal Abstract: Program security analysis has been studied for decades. Various techniques, such as fuzzing, taint analysis, symbolic execution, have demonstrated their successes in vulnerability assessment. Today, the availability of a large amount of program semantic data (e.g., manuals, developer documentation, related web content), and the advance of artificial intelligence technologies make it increasingly feasible to simulate human intelligence in understanding program semantics to discover software vulnerability automatically. In this talk, I will discuss my research toward in-depth and systematic semantic supports for automatic vulnerability assessment. Particularly, I will focus on two systems — Advance and Dilution — which automatically analyzes the developer’s guide to infer potential security flaws and API misuse, respectively. Bio: Xiaojing Liao is an Assistant Professor in the Department of Computer Science at Indiana University Bloomington. Her research interests include data-driven security and privacy, with specific focuses on system security, cybercrime, as well as cyber-physical systems security and privacy. She has published papers on leading system security venues such as S&P (Oakland), Usenix Security, CCS, and NDSS. She is the recipient of the ACM SIGSAC Dissertations Award and NDSS Distinguish Paper Award.

1 0

CS Colloquium Speaker: Amy Zhang, Monday, March 8 at 4:30
by Emily C. Lawrence 02 Mar '21

02 Mar '21

CS Colloquium Speaker Amy Zhang, McGill University and the Mila Institute Date: Monday, March 8 Time: 4:30pm EST Host: Tom Griffiths Event page: [ https://www.cs.princeton.edu/events/26041 | https://www.cs.princeton.edu/events/26041 ] Register for Zoom webinar here: [ https://princeton.zoom.us/webinar/register/WN_yb-TRAQJSwyBTo7RSDfhUA | https://princeton.zoom.us/webinar/register/WN_yb-TRAQJSwyBTo7RSDfhUA ] Title: Exploiting Latent Structure and Bisimulation Metrics for Better Generalization in Reinforcement Learning Abstract: The advent of deep learning has shepherded unprecedented progress in various fields of machine learning. Despite recent advances in deep reinforcement learning (RL) algorithms, however, there is no method today that exhibits anywhere near the generalization that we have seen in computer vision and NLP. Indeed, one might ask whether deep RL algorithms are even capable of the kind of generalization that is needed for open-world environments. This challenge is fundamental and will not be solved with incremental algorithmic advances. In this talk, we propose to incorporate different assumptions that better reflect the real world and allow the design of novel algorithms with theoretical guarantees to address this fundamental problem. We first present how state abstractions can accelerate reinforcement learning from rich observations, such as images, without relying either on domain knowledge or pixel-reconstruction. Our goal is to learn state abstractions that both provide for effective downstream control and invariance to task-irrelevant details. We use bisimulation metrics to quantify behavioral similarity between states, and learn robust latent representations which encode only the task-relevant information from observations. We provide theoretical guarantees for the learned approximate abstraction and extend this notion to families of tasks with varying dynamics. Bio: I am a final year PhD candidate at McGill University and the Mila Institute, co-supervised by Profs. Joelle Pineau and Doina Precup. I am also a researcher at Facebook AI Research. My work focuses on bridging theory and practice through learning approximate state abstractions and learning representations for generalization in reinforcement learning. I previously obtained an M.Eng. in EECS and dual B.Sci. degrees in Mathematics and EECS from MIT.

1 0

CS Colloquium Speaker: Sanmi Koyejo, today at 12:30
by Emily C. Lawrence 01 Mar '21

01 Mar '21

CS Colloquium Speaker Sanmi Koyejo, University of Illinois at Urbana-Champaign Host: Barbara Engelhardt Date: Monday, March 1 Time: 12:30pm EST Event page: https://www.cs.princeton.edu/events/26040 Register for Zoom webinar here: https://princeton.zoom.us/webinar/register/WN_XMi3YQUVRn20UGAhEncHpQ Title: The Measurement and Mismeasurement of Trustworthy ML Abstract: Across healthcare, science, and engineering, we increasingly employ machine learning (ML) to automate decision-making that, in turn, affects our lives in profound ways. However, ML can fail, with significant and long-lasting consequences. Reliably measuring such failures is the first step towards building robust and trustworthy learning machines. Consider algorithmic fairness, where widely-deployed fairness metrics can exacerbate group disparities and result in discriminatory outcomes. Moreover, existing metrics are often incompatible. Hence, selecting fairness metrics is an open problem. Measurement is also crucial for robustness, particularly in federated learning with error-prone devices. Here, once again, models constructed using well-accepted robustness metrics can fail. Across ML applications, the dire consequences of mismeasurement are a recurring theme. This talk will outline emerging strategies for addressing the measurement gap in ML and how this impacts trustworthiness. Bio: Sanmi (Oluwasanmi) Koyejo is an Assistant Professor in the Department of Computer Science at the University of Illinois at Urbana-Champaign. Koyejo's research interests are in developing the principles and practice of trustworthy machine learning. Additionally, Koyejo focuses on applications to neuroscience and healthcare. Koyejo has been the recipient of several awards, including a best paper award from the conference on uncertainty in artificial intelligence (UAI), a Kavli Fellowship, an IJCAI early career spotlight, and a trainee award from the Organization for Human Brain Mapping (OHBM). Koyejo serves on the board of the Black in AI organization.

1 0

Reminder: Sanmi Koyejo, tomorrow at 12:30
by Emily C. Lawrence 28 Feb '21

28 Feb '21

CS Colloquium Speaker Sanmi Koyejo, University of Illinois at Urbana-Champaign Host: Barbara Engelhardt Date: Monday, March 1 Time: 12:30pm EST Event page: https://www.cs.princeton.edu/events/26040 Register for Zoom webinar here: https://princeton.zoom.us/webinar/register/WN_XMi3YQUVRn20UGAhEncHpQ Title: The Measurement and Mismeasurement of Trustworthy ML Abstract: Across healthcare, science, and engineering, we increasingly employ machine learning (ML) to automate decision-making that, in turn, affects our lives in profound ways. However, ML can fail, with significant and long-lasting consequences. Reliably measuring such failures is the first step towards building robust and trustworthy learning machines. Consider algorithmic fairness, where widely-deployed fairness metrics can exacerbate group disparities and result in discriminatory outcomes. Moreover, existing metrics are often incompatible. Hence, selecting fairness metrics is an open problem. Measurement is also crucial for robustness, particularly in federated learning with error-prone devices. Here, once again, models constructed using well-accepted robustness metrics can fail. Across ML applications, the dire consequences of mismeasurement are a recurring theme. This talk will outline emerging strategies for addressing the measurement gap in ML and how this impacts trustworthiness. Bio: Sanmi (Oluwasanmi) Koyejo is an Assistant Professor in the Department of Computer Science at the University of Illinois at Urbana-Champaign. Koyejo's research interests are in developing the principles and practice of trustworthy machine learning. Additionally, Koyejo focuses on applications to neuroscience and healthcare. Koyejo has been the recipient of several awards, including a best paper award from the conference on uncertainty in artificial intelligence (UAI), a Kavli Fellowship, an IJCAI early career spotlight, and a trainee award from the Organization for Human Brain Mapping (OHBM). Koyejo serves on the board of the Black in AI organization.

1 0

CS Colloquium Speaker: Huacheng Yu, Thursday, March 4 at 12:30
by Emily C. Lawrence 25 Feb '21

25 Feb '21

CS Colloquium Speaker Huacheng Yu, Princeton University Host: Zeev Dvir Date: Thursday, March 4 Time: 12:30pm EST Event page: [ https://www.cs.princeton.edu/events/26051 | https://www.cs.princeton.edu/events/26051 ] Register for Zoom webinar here: [ https://princeton.zoom.us/webinar/register/WN_pNvnCKFHS_qoXLEEOSMOwA | https://princeton.zoom.us/webinar/register/WN_pNvnCKFHS_qoXLEEOSMOwA ] Title: Data Structures and Algorithms in Sublinear Computation Abstract: Sublinear algorithms are sustainable under the exponential increase of data volume and processing speed. Such algorithms use a sublinear amount of resources, e.g., spending time, space, or communication that is asymptotically smaller than the input data size. Typical examples include data structures, which compute query functions of the data in sublinear time, and streaming algorithms, which make one pass over massive data streams while maintaining a sublinear-sized memory. In this talk, I will give an overview of my work in sublinear computation, focusing on succinct data structures and distributed graph sketching algorithms. I will first discuss my work on a nearly optimal data structure for the dictionary problem, for which the textbook solution uses hash tables. Then, I will talk about detecting the connectivity of graphs using distributed sketching, and my recent work showing the optimality of a well-known sketching algorithm (the AGM sketch). I will conclude the talk with discussion on future directions and my other work in theoretical computer science. Bio: Huacheng Yu is an associate research scholar in the Department of Computer Science at Princeton University. His research interests include data structures and streaming algorithms, and other directions in theoretical computer science such as communication complexity and graph algorithms. Prior to Princeton, Huacheng was a postdoctoral researcher at Harvard University hosted by Jelani Nelson and Madhu Sudan. He received his Ph.D. from Stanford University (advised by Ryan Williams and Omer Reingold) and B.Eng from Tsinghua University, both in Computer Science.

1 0

CS Colloquium Speaker: Sanmi Koyejo, Monday, Mar 1 at 12:30
by Emily C. Lawrence 23 Feb '21

23 Feb '21

CS Colloquium Speaker Sanmi Koyejo, University of Illinois at Urbana-Champaign Host: Barbara Engelhardt Date: Monday, March 1 Time: 12:30pm EST Event page: https://www.cs.princeton.edu/events/26040 Register for Zoom webinar here: https://princeton.zoom.us/webinar/register/WN_XMi3YQUVRn20UGAhEncHpQ Title: The Measurement and Mismeasurement of Trustworthy ML Abstract: Across healthcare, science, and engineering, we increasingly employ machine learning (ML) to automate decision-making that, in turn, affects our lives in profound ways. However, ML can fail, with significant and long-lasting consequences. Reliably measuring such failures is the first step towards building robust and trustworthy learning machines. Consider algorithmic fairness, where widely-deployed fairness metrics can exacerbate group disparities and result in discriminatory outcomes. Moreover, existing metrics are often incompatible. Hence, selecting fairness metrics is an open problem. Measurement is also crucial for robustness, particularly in federated learning with error-prone devices. Here, once again, models constructed using well-accepted robustness metrics can fail. Across ML applications, the dire consequences of mismeasurement are a recurring theme. This talk will outline emerging strategies for addressing the measurement gap in ML and how this impacts trustworthiness. Bio: Sanmi (Oluwasanmi) Koyejo is an Assistant Professor in the Department of Computer Science at the University of Illinois at Urbana-Champaign. Koyejo's research interests are in developing the principles and practice of trustworthy machine learning. Additionally, Koyejo focuses on applications to neuroscience and healthcare. Koyejo has been the recipient of several awards, including a best paper award from the conference on uncertainty in artificial intelligence (UAI), a Kavli Fellowship, an IJCAI early career spotlight, and a trainee award from the Organization for Human Brain Mapping (OHBM). Koyejo serves on the board of the Black in AI organization.

1 0

4:30pm Tue Mar 9 talk on "Building secure distributed systems atop the insecure Internet"
by Jennifer Rexford 22 Feb '21

22 Feb '21

> > > > ECE SEMINAR > > Speaker: > Maria Apostolaki, ETH > Title: > Building secure distributed systems atop the insecure Internet > Day: > Tuesday, March 9, 2021 > Time: > 4:30 pm > Room: > Please register here <https://princeton.zoom.us/meeting/register/tJUlc-2oqTwiHdZVqHbVqzMkeIX1y6v7…> > Host: > Prateek Mittal > > Abstract: > Distributed systems are increasingly important for our everyday life, allowing for high performance, fault tolerance, and flexibility. Many of these systems nowadays rely on the inherently insecure Internet infrastructure. Surely though, they should have been designed to take this into account...? > In this talk, I will answer negatively to this question using a concrete example: public blockchain systems such as Bitcoin. These are novel distributed systems that are designed according to stringent failure models. In this context, I will explain how an adversary controlling pieces of Internet infrastructure can practically compromise: (i) Bitcoin's consensus protocol (by partitioning the network); (ii) Bitcoin's anonymity guarantees (by mapping pseudonyms to real-world identities); and (iii) Bitcoin's availability (by eclipsing clients). > While these attacks are worrying, I will also introduce practical and effective defenses to counter them both at the network and the application layer. Beyond Bitcoin, this work teaches essential lessons for distributed-system design. > > > > Bio: > Maria Apostolaki is a PhD Student at ETH Zurich advised by Laurent Vanbever. During her studies, she has been a visiting student at MIT (2019) and a research intern at Microsoft Research (2018) and Google (2017). Before joining ETH, she earned her diploma in Electrical and Computer Engineering at the National Technical University of Athens, Greece. > Her research focuses on building secure, performant, and deployable networked systems using both hardware and software. Her research led to discovering significant vulnerabilities in the Bitcoin system and spearheaded changes to the Bitcoin codebase. Maria’s work received widespread media coverage and was awarded an Applied Networking Research Prize by IETF.

1 0

March 1, 2021 - ECE Circuits Seminar - Carmel Majidi, Carnegie Mellon University
by Emily C. Lawrence 22 Feb '21

22 Feb '21

ECE CIRCUITS SEMINAR Speaker: Carmel Majidi , Carnegie Mellon University Title: Soft-Matter Engineering for Robotics and Wearables Day: Monday, March 1, 2021 Time: 4:30 pm Zoom Link: Meeting link below Host: Prof. Kaushik Sengupta Abstract: Progress in soft lithography and soft materials integration have led to extraordinary new classes of soft-matter sensors, circuits, and transducers. These material technologies are composed almost entirely out of soft matter – elastomers, gels, and conductive fluids like eutectic gallium-indium (EGaIn) – and represent the building blocks for machines and electronics that are soft, flexible, and stretchable. Because of their intrinsic compliance and elasticity, such devices can be incorporated into soft, biologically-inspired robots or be worn on the body and operate continuously without impairing natural body motion. In this talk, I will review recent contributions from my research group in creating soft multifunctional materials for wearable electronics and soft robotics using these emerging practices in “soft-matter engineering.” In particular, I will focus on elastomer composites and microfluidic EGaIn architectures for highly stretchable digital electronics, wearable energy harvesting, and electrically-responsive actuation. When possible, I will relate the design and operation of these soft-matter technologies to underlying principles of soft matter physics and practices in controls and machine. In addition to presenting my own research in the field, I will also briefly review broader efforts and emerging challenges in utilizing soft electronic materials for applications in wearable electronics and soft robotics . Bio: Carmel Majidi is the Clarence H. Adamson Professor of Mechanical Engineering at Carnegie Mellon University, where he leads the Soft Machines Lab. His lab is dedicated to the discovery of novel material architectures that allow machines and electronics to be soft, elastically deformable, and biomechanically compatible. Currently, his research is focused on fluid-filled elastomers that exhibit unique combinations of mechanical, electrical, and thermal properties and can function as “artificial” skin, nervous tissue, and muscle for soft robotics and wearables. Carmel has received grants from industry and federal agencies along with early career awards from DARPA, ONR, AFOSR, and NASA to explore challenges in soft-matter engineering and robotics. Prior to arriving at CMU, Prof. Majidi had postdoctoral appointments at Harvard and Princeton Universities and received his PhD in Electrical Engineering at UC Berkeley. Join Zoom Meeting : [ https://princeton.zoom.us/j/99499632211?pwd=Qm4wWFFWSUFyMjJVQ3lseFh1TGNyUT09 | https://princeton.zoom.us/j/99499632211?pwd=Qm4wWFFWSUFyMjJVQ3lseFh1TGNyUT09 ]

1 0

CS Colloquium Speaker: Ellen Vitercik today at 12:30pm EST
by Emily C. Lawrence 22 Feb '21

22 Feb '21

CS Colloquium Speaker Ellen Vitercik, Carnegie Mellon University Host: Sanjeev Arora Date: Monday, February 22 Time: 12:30pm EST Event page: [ https://www.cs.princeton.edu/events/26042 | https://www.cs.princeton.edu/events/26042 ] Register for the Zoom webinar here: [ https://princeton.zoom.us/webinar/register/WN_gbMN0OUJQRainEiRTS4VYQ | https://princeton.zoom.us/webinar/register/WN_gbMN0OUJQRainEiRTS4VYQ ] Title: Integrating Machine Learning into Algorithm Design Abstract: An important property of those algorithms that are typically used in practice is broad applicability—the ability to solve problems across diverse domains. However, the default, out-of-the-box performance of these algorithms can be unsatisfactory, with slow runtime, poor solution quality, and even negative long-term social ramifications. In practice, there is often ample data available about the types of problems that an algorithm will be run on, data that can potentially be harnessed to fine-tune the algorithm’s performance. We therefore need principled approaches for using this data to obtain strong application-specific performance guarantees. In this talk, I will give an overview of my research that provides practical methods built on firm theoretical foundations for incorporating machine learning and optimization into the process of algorithm design, selection, and configuration. I will describe my contributions across several diverse domains, including integer programming, clustering, mechanism design, and computational biology. As I will demonstrate, these seemingly disparate areas are connected by overarching structure which implies broadly-applicable guarantees. Bio: Ellen Vitercik is a PhD student at Carnegie Mellon University where she is co-advised by Maria-Florina Balcan and Tuomas Sandholm. Her research revolves around artificial intelligence, algorithm design, and the interface between economics and computation, with a particular focus on machine learning theory. Among other honors, she is a recipient of the Exemplary Artificial Intelligence Track Paper Award at EC'19, the Best Presentation by a Student or Postdoctoral Researcher Award at EC'19, the NSF Graduate Research Fellowship, the IBM PhD Fellowship, the Fellowship in Digital Health from CMU's Center for Machine Learning and Health, and the Teaching Assistant of the Year Award from CMU's Machine Learning Department. This talk will be recorded.

1 0