CS Department Colloquium Series: week of Mar 27-31
Here are next week's CS Department Colloquium Series talks. As always, you can find the full calendar of events here: https://www.cs.princeton.edu/general/newsevents/events Speaker: Adam Fisch, Massachusetts Institute of Technology Date: Monday, March 27 Time: 12:30pm EST Location: CS 105 Host: Karthik Narasimhan Event page: https://www.cs.princeton.edu/events/26363 Title: Towards Efficient and Reliable Machine Learning for Natural Language Processing (and Beyond) Abstract: In this talk, I will introduce work on fundamental techniques for building and deploying effective natural language processing (NLP) systems that are also efficient and reliable. Specifically, I will address three interconnected challenges for modern machine learning in NLP: how to quickly adapt foundation models to new tasks with limited data, how to dynamically reconfigure large architectures for more efficient computation, and how to develop powerful theoretical tools for rigorous, yet practical, uncertainty quantification. To conclude, I will highlight a number of my future research directions, as well as extensions to interesting applications beyond natural language. Bio: Adam Fisch is a PhD candidate at MIT working with Regina Barzilay and Tommi Jaakkola, and a recipient of an NSF Graduate Research Fellowship. His research centers around principled methods for efficient and reliable machine learning systems that work effectively in realistic scenarios, and has appeared in top-tier venues such as *ACL, ICLR, ICML, and NeurIPS. Adam also served as a co-instructor for the tutorial on Uncertainty Estimation for NLP at COLING 2022, and as a co-organizer of the Machine Reading for Question Answering workshops at EMNLP 2019 and 2021. Prior to MIT, Adam was a research engineer at Meta (Facebook) AI Research for two years, and studied mechanical engineering as an undergraduate at Princeton University. Speaker: Aviral Kumar, University of California, Berkeley Date: Tuesday, March 28 Time: 12:30pm EST Location: CS 105 Host: Jia Deng Event page: https://www.cs.princeton.edu/events/26377 Title: Reinforcement Learning from Static Datasets: Algorithms, Analysis and Applications Abstract: Typically, reinforcement learning (RL) methods rely on trial-and-error interaction with the environment from scratch to discover effective behaviors. While this sort of paradigm has the potential to discover good strategies, this paradigm also inhibits RL methods from collecting enough experience or training data in real-world problems where active interaction is expensive (e.g., in drug design) or dangerous (e.g., for robots operating around humans). My work develops approaches to alleviate this limitation: how can we learn policies to effectively make decisions entirely from previously-collected, static datasets in an offline manner? In this talk, I will discuss challenges that appear in this kind of offline reinforcement learning (offline RL) and develop algorithms and techniques to address these challenges. I will then discuss how my approaches for offline RL and decision-making have enabled us to make progress in real-world problems such as hardware accelerator design, robotic manipulation, and computational chemistry. Finally, I will discuss how we can enable offline RL methods to benefit from generalization capabilities offered by large and expressive models, similar to supervised learning. Bio: Aviral Kumar is a final year Ph.D. student at UC Berkeley. His research focuses on developing effective and reliable approaches for (sequential) decision-making. Towards this goal, he focuses on designing reinforcement learning techniques to static datasets and on understanding and applying these methods in practice. Before his Ph.D., Aviral obtained his B.Tech. in Computer Science from IIT Bombay in India. He is a recipient of the C.V. & Daulat Ramamoorthy Distinguished Research Award, given to 1 PhD student in EECS at Berkeley for outstanding contributions to a new area of research in computer science, Facebook Ph.D. Fellowship in Machine Learning and Apple Scholars in AI/ML Ph.D. Fellowship.
Here are next week's CS Department Colloquium Series talks. As always, you can find the full calendar of events here: https://www.cs.princeton.edu/general/newsevents/events Speaker: Venkat Arun, Massachusetts Institute of Technology Date: Monday, April 3 Time: 12:30pm EST Location: CS 105 Host: Ravi Netravali Event page: https://www.cs.princeton.edu/events/26346 Title: Designing Provably Performant Networked Systems Abstract: As networked systems become critical infrastructure, their design must reflect their new societal role. Today, we build systems with hundreds of heuristics but often do not understand their inherent and emergent behaviors. I will present a set of tools and techniques to prove performance properties of heuristics running in real-world conditions. Rigorous proofs can not only inspire confidence in our designs, but also give counter-intuitive insights about their performance. A key theme in our approach is to model uncertainty in systems using non-random, non-deterministic objects that cover a wide range of possible behaviors under a single abstraction. Such models allow us to analyze complex system behaviors using automated reasoning techniques. I will present automated tools to analyze congestion control and process scheduling algorithms. These tools prove performance properties and find counter-examples where widely deployed heuristics fail. I will also show that current end-to-end congestion control algorithms that bound delay cannot avoid starvation and present a method to beamform wireless signals using thousands of antennas. Bio: Venkat Arun is a PhD candidate at MIT working with Hari Balakrishnan and Mohammad Alizadeh. His work spans internet congestion control, video streaming, privacy-preserving computation, wireless networks, and mobile systems. Across these areas, a unifying theme of his work is to bridge between heuristics that systems use in practice and proofs of how well they work. He believes that rigorous proof combined with automated reasoning will enable us to make networked systems more robust and performant. He has won two ACM SIGCOMM best paper awards and the president of India gold medal. Speaker: Rika Antonova, Stanford University Date: Tuesday, April 4 Time: 12:30pm EST Location: CS 105 Host: Szymon Rusinkiewicz Event page: https://www.cs.princeton.edu/events/26378 Title: Enabling Self-Sufficient Robot Learning Abstract: Autonomous exploration and data-efficient learning are important ingredients for helping machine learning handle the complexity and variety of real-world interactions. In this talk, I will describe methods that provide these ingredients and serve as building blocks for enabling self-sufficient robot learning. First, I will outline a family of methods that facilitate active global exploration. Specifically, they enable ultra data-efficient Bayesian optimization in reality by leveraging experience from simulation to shape the space of decisions. In robotics, these methods enable success with a budget of only 10-20 real robot trials for a range of tasks: bipedal and hexapod walking, task-oriented grasping, and nonprehensile manipulation. Next, I will describe how to bring simulations closer to reality. This is especially important for scenarios with highly deformable objects, where simulation parameters influence the dynamics in unintuitive ways. The success here hinges on finding a good representation for the state of deformables. I will describe adaptive distribution embeddings that provide an effective way to incorporate noisy state observations into modern Bayesian tools for simulation parameter inference. This novel representation ensures success in estimating posterior distributions over simulation parameters, such as elasticity, friction, and scale, even for scenarios with highly deformable objects and using only a small set of real-world trajectories. Lastly, I will share a vision of using distribution embeddings to make the space of stochastic policies in reinforcement learning suitable for global optimization. This research direction involves formalizing and learning novel distance metrics on this space and will support principled ways of seeking diverse behaviors. This can unlock truly autonomous learning, where learning agents have incentives to explore, build useful internal representations and discover a variety of effective ways of interacting with the world. Bio: Rika is a postdoctoral scholar at Stanford University and a recipient of the NSF/CRA Computing Innovation Fellowship. Rika completed her Ph.D. work on data-efficient simulation-to-reality transfer at KTH. Earlier, she obtained a research Master's degree from the Robotics Institute at Carnegie Mellon University. Before that, Rika was a software engineer at Google, first in the Search Personalization group, then in the Character Recognition team (developing open-source OCR engine Tesseract). Speaker: Lydia Liu, Cornell University Date: Wednesday, April 5 Time: 12:30pm EST Location: CS 105 Hosts: Olga Russakovsky and Peter Ramadge Event page: https://www.cs.princeton.edu/events/26353 Title: Towards Responsible Machine Learning in Societal Systems Abstract: Machine learning systems are deployed in consequential domains such as education, employment, and credit, where decisions have profound effects on socioeconomic opportunity and life outcomes. High stakes decision settings present new statistical, algorithmic, and ethical challenges. In this talk, we examine the distributive impact of machine learning algorithms in societal contexts, and investigate the algorithmic and sociotechnical interventions that bring machine learning systems into alignment with societal values---equity and long-term welfare. First, we study the dynamic interactions between machine learning algorithms and populations, for the purpose of mitigating disparate impact in applications such as algorithmic lending and hiring. Next, we consider data-driven decision systems in competitive environments such as markets, and devise learning algorithms to ensure efficiency and allocative fairness. We end by outlining future directions for responsible machine learning in societal systems that bridge the gap between the optimization of predictive models and the evaluation of downstream decisions and impact. Bio: Lydia T. Liu is a postdoctoral researcher in Computer Science at Cornell University, working with Jon Kleinberg, Karen Levy, and Solon Barocas. Her research examines the theoretical foundations of machine learning and algorithmic decision-making, with a focus on societal impact and human welfare. She obtained her PhD in Electrical Engineering and Computer Sciences from UC Berkeley, advised by Moritz Hardt and Michael Jordan, and has received a Microsoft Ada Lovelace Fellowship, an Open Philanthropy AI Fellowship, an NUS Development Grant, and a Best Paper Award at the International Conference on Machine Learning. This talk is co-sponsored with Electrical and Computer Engineering and the Center for Information Technology Policy. Speaker: Mae Milano, University of California, Berkeley Date: Monday, March 6 Time: 12:30pm EST Location: CS 105 Host: Andrew Appel Event page: https://www.cs.princeton.edu/events/26344 Title: Programming Distributed Systems Abstract: Our interconnected world is increasingly reliant on distributed systems of unprecedented scale, serving applications which must share state across the globe. And, despite decades of research, we're still not sure how to program them! In this talk, I'll show how to use ideas from programming languages to make programming at scale easier, without sacrificing performance, correctness, or expressive power in the process. We'll see how slight tweaks to modern imperative programming languages can provably eliminate common errors due to replica consistency or concurrency---with little to no programmer effort. We'll see how new language designs can unlock new systems designs, yielding both more comprehensible protocols and better performance. And we'll conclude by imagining together the role that a new cloud-centric programming language could play in the next generation of distributed programs. Bio: Mae Milano is a postdoctoral scholar at UC Berkeley working at the intersection of Programming Languages, Distributed Systems, and Databases. Her work has appeared at top-tier venues including PLDI, OOPSLA, POPL, VLDB, and TOCS, and has attracted the attention of the Swift language team. She is a recipient of the NDSEG Fellowship, has won several awards for her writing and service, and is a founding member of the Computing Connections Fellowship's selection committee (https://computingconnections.org/).
Correction: Speaker Mae Milano will be on Thursday, April 6 . Speaker: Venkat Arun, Massachusetts Institute of Technology Date: Monday, April 3 Time: 12:30pm EST Location: CS 105 Host: Ravi Netravali Event page: https://www.cs.princeton.edu/events/26346 Title: Designing Provably Performant Networked Systems Abstract: As networked systems become critical infrastructure, their design must reflect their new societal role. Today, we build systems with hundreds of heuristics but often do not understand their inherent and emergent behaviors. I will present a set of tools and techniques to prove performance properties of heuristics running in real-world conditions. Rigorous proofs can not only inspire confidence in our designs, but also give counter-intuitive insights about their performance. A key theme in our approach is to model uncertainty in systems using non-random, non-deterministic objects that cover a wide range of possible behaviors under a single abstraction. Such models allow us to analyze complex system behaviors using automated reasoning techniques. I will present automated tools to analyze congestion control and process scheduling algorithms. These tools prove performance properties and find counter-examples where widely deployed heuristics fail. I will also show that current end-to-end congestion control algorithms that bound delay cannot avoid starvation and present a method to beamform wireless signals using thousands of antennas. Bio: Venkat Arun is a PhD candidate at MIT working with Hari Balakrishnan and Mohammad Alizadeh. His work spans internet congestion control, video streaming, privacy-preserving computation, wireless networks, and mobile systems. Across these areas, a unifying theme of his work is to bridge between heuristics that systems use in practice and proofs of how well they work. He believes that rigorous proof combined with automated reasoning will enable us to make networked systems more robust and performant. He has won two ACM SIGCOMM best paper awards and the president of India gold medal. Speaker: Rika Antonova, Stanford University Date: Tuesday, April 4 Time: 12:30pm EST Location: CS 105 Host: Szymon Rusinkiewicz Event page: https://www.cs.princeton.edu/events/26378 Title: Enabling Self-Sufficient Robot Learning Abstract: Autonomous exploration and data-efficient learning are important ingredients for helping machine learning handle the complexity and variety of real-world interactions. In this talk, I will describe methods that provide these ingredients and serve as building blocks for enabling self-sufficient robot learning. First, I will outline a family of methods that facilitate active global exploration. Specifically, they enable ultra data-efficient Bayesian optimization in reality by leveraging experience from simulation to shape the space of decisions. In robotics, these methods enable success with a budget of only 10-20 real robot trials for a range of tasks: bipedal and hexapod walking, task-oriented grasping, and nonprehensile manipulation. Next, I will describe how to bring simulations closer to reality. This is especially important for scenarios with highly deformable objects, where simulation parameters influence the dynamics in unintuitive ways. The success here hinges on finding a good representation for the state of deformables. I will describe adaptive distribution embeddings that provide an effective way to incorporate noisy state observations into modern Bayesian tools for simulation parameter inference. This novel representation ensures success in estimating posterior distributions over simulation parameters, such as elasticity, friction, and scale, even for scenarios with highly deformable objects and using only a small set of real-world trajectories. Lastly, I will share a vision of using distribution embeddings to make the space of stochastic policies in reinforcement learning suitable for global optimization. This research direction involves formalizing and learning novel distance metrics on this space and will support principled ways of seeking diverse behaviors. This can unlock truly autonomous learning, where learning agents have incentives to explore, build useful internal representations and discover a variety of effective ways of interacting with the world. Bio: Rika is a postdoctoral scholar at Stanford University and a recipient of the NSF/CRA Computing Innovation Fellowship. Rika completed her Ph.D. work on data-efficient simulation-to-reality transfer at KTH. Earlier, she obtained a research Master's degree from the Robotics Institute at Carnegie Mellon University. Before that, Rika was a software engineer at Google, first in the Search Personalization group, then in the Character Recognition team (developing open-source OCR engine Tesseract). Speaker: Lydia Liu, Cornell University Date: Wednesday, April 5 Time: 12:30pm EST Location: CS 105 Hosts: Olga Russakovsky and Peter Ramadge Event page: https://www.cs.princeton.edu/events/26353 Title: Towards Responsible Machine Learning in Societal Systems Abstract: Machine learning systems are deployed in consequential domains such as education, employment, and credit, where decisions have profound effects on socioeconomic opportunity and life outcomes. High stakes decision settings present new statistical, algorithmic, and ethical challenges. In this talk, we examine the distributive impact of machine learning algorithms in societal contexts, and investigate the algorithmic and sociotechnical interventions that bring machine learning systems into alignment with societal values---equity and long-term welfare. First, we study the dynamic interactions between machine learning algorithms and populations, for the purpose of mitigating disparate impact in applications such as algorithmic lending and hiring. Next, we consider data-driven decision systems in competitive environments such as markets, and devise learning algorithms to ensure efficiency and allocative fairness. We end by outlining future directions for responsible machine learning in societal systems that bridge the gap between the optimization of predictive models and the evaluation of downstream decisions and impact. Bio: Lydia T. Liu is a postdoctoral researcher in Computer Science at Cornell University, working with Jon Kleinberg, Karen Levy, and Solon Barocas. Her research examines the theoretical foundations of machine learning and algorithmic decision-making, with a focus on societal impact and human welfare. She obtained her PhD in Electrical Engineering and Computer Sciences from UC Berkeley, advised by Moritz Hardt and Michael Jordan, and has received a Microsoft Ada Lovelace Fellowship, an Open Philanthropy AI Fellowship, an NUS Development Grant, and a Best Paper Award at the International Conference on Machine Learning. This talk is co-sponsored with Electrical and Computer Engineering and the Center for Information Technology Policy. Speaker: Mae Milano, University of California, Berkeley Date: Thursday, April 6 Time: 12:30pm EST Location: CS 105 Host: Andrew Appel Event page: https://www.cs.princeton.edu/events/26344 Title: Programming Distributed Systems Abstract: Our interconnected world is increasingly reliant on distributed systems of unprecedented scale, serving applications which must share state across the globe. And, despite decades of research, we're still not sure how to program them! In this talk, I'll show how to use ideas from programming languages to make programming at scale easier, without sacrificing performance, correctness, or expressive power in the process. We'll see how slight tweaks to modern imperative programming languages can provably eliminate common errors due to replica consistency or concurrency---with little to no programmer effort. We'll see how new language designs can unlock new systems designs, yielding both more comprehensible protocols and better performance. And we'll conclude by imagining together the role that a new cloud-centric programming language could play in the next generation of distributed programs. Bio: Mae Milano is a postdoctoral scholar at UC Berkeley working at the intersection of Programming Languages, Distributed Systems, and Databases. Her work has appeared at top-tier venues including PLDI, OOPSLA, POPL, VLDB, and TOCS, and has attracted the attention of the Swift language team. She is a recipient of the NDSEG Fellowship, has won several awards for her writing and service, and is a founding member of the Computing Connections Fellowship's selection committee (https://computingconnections.org/). _______________________________________________ talks mailing list talks@lists.cs.princeton.edu To edit subscription settings or remove yourself, use this link: https://lists.cs.princeton.edu/mailman/listinfo/talks
Here are next week's CS Department Colloquium Series talks. As always, you can find the full calendar of events here: https://www.cs.princeton.edu/general/newsevents/events Speaker: Yang Liu, Stanford University Date: Monday, April 10 Time: 12:30pm EST Location: CS 105 Host: Gillat Kol Event page: https://www.cs.princeton.edu/events/26400 Title: Graphs, Optimization, Geometry, and Fast Algorithms Abstract: Discrete combinatorial structures such as graphs and Boolean matrices are prevalent in modern computation. The massive size of modern data motivates the design of efficient algorithms for processing these combinatorial datasets. In this talk, I will describe how to use techniques from continuous optimization and geometry to gain insights into the structure of problems in these combinatorial settings. Using these insights, I will present new efficient algorithms for several fundamental problems at the intersection of combinatorial algorithms, continuous optimization, and high-dimensional geometry, including maximum flows in almost linear time, discrepancy minimization, and linear regression. We conclude by discussing the exciting new lines of research and open problems that these techniques have opened up. Bio: Yang P. Liu is a fifth year PhD student at Stanford advised by Aaron Sidford. He completed his undergraduate studies at MIT in 2018. He has broad interests in computer science, and his research focuses on the design of efficient algorithms based on graph theory, convex optimization, and high-dimensional geometry. His work has been recognized by ITCS and STOC best student paper awards, and a FOCS best paper award. CITP & CS Colloquium Speaker Speaker: Amanda Coston, Carnegie Mellon University Date: Tuesday, April 11 Time: 12:30pm EST Location: CS 105 Host: Aleksandra Korolova Event page: https://www.cs.princeton.edu/events/26375 Title: Responsible Machine Learning through the Lens of Causal Inference Abstract: Machine learning algorithms are widely used for decision-making in societally high-stakes settings from child welfare and criminal justice to healthcare and consumer lending. Recent history has illuminated numerous examples where these algorithms proved unreliable or inequitable. In this talk I show how causal inference enables us to more reliably evaluate such algorithms’ performance and equity implications. In the first part of the talk, I demonstrate that standard evaluation procedures fail to address missing data and as a result, often produce invalid assessments of algorithmic performance. I propose a new evaluation framework that addresses missing data by using counterfactual techniques to estimate unknown outcomes. Using this framework, I propose counterfactual analogues of common predictive performance and algorithmic fairness metrics that are tailored to decision-making settings. I provide double machine learning-style estimators for these metrics that achieve fast rates & asymptotic normality under flexible nonparametric conditions. I present empirical results in the child welfare setting using data from Allegheny County’s Department of Human Services. In the second half of the talk, I propose novel causal inference methods to audit for bias in key decision points in contexts where machine learning algorithms are used. A common challenge is that data about decisions are often observed under outcome-dependent sampling. I develop a counterfactual audit for biased decision-making in settings with outcome-dependent data. Using data from the Stanford Open Policing Project, I demonstrate how this method can identify racial bias in the most common entry point to the criminal justice system: police traffic stops. To conclude, I situate my work in the broader question of governance in responsible machine learning. Bio: Amanda Coston is a PhD student in Machine Learning and Public Policy at Carnegie Mellon University. Her research investigates how to make algorithmic decision-making more reliable and more equitable using causal inference and machine learning. Prior to her PhD, she worked at Microsoft, the consultancy Teneo, and the Nairobi-based startup HiviSasa. She earned a B.S.E from Princeton in computer science with a certificate in public policy. Amanda is a Meta Research PhD Fellow, K & L Gates Presidential Fellow in Ethics and Computational Technologies, and NSF GRFP Fellow, and has received several Rising Star honors. This seminar is cosponsored by the Center for Information Technology Policy and the department of Computer Science. Speaker: Pravesh Kothari, Carnegie Mellon University Date: Thursday, April 13 Time: 12:30pm EST Location: CS 105 Host: Mark Braverman Event page: https://www.cs.princeton.edu/events/26390 Title: Towards a Unified Approach to Average-Case Algorithm Design Abstract: Solving non-convex optimization problems on probabilistic models of inputs lies at the heart of foundational algorithmic challenges arising in high-dimensional statistical data analysis, beyond-worst-case combinatorial optimization, cryptography, and statistical physics. In this talk, I will present a new method for average-case algorithm design that relies on a concrete polynomial time meta-algorithm called the sum-of-squares method. This method yields substantially improved and often nearly optimal guarantees for a wide range of problems. I will focus on the impact of this method on two prominent areas of average-case algorithm design: 1) High-dimensional statistical estimation, where this method has led to efficient algorithms for classical data analysis tasks that provably tolerate adversarial data corruption while incurring minimal possible error. The resulting applications range from new robust estimators in high dimensions for basic tasks such as computing mean, covariance, and moments of data to more sophisticated tasks such as regression, clustering, sparse recovery, and fitting mixture models. Most recently, this theory led to the first efficient algorithm for robustly learning a high-dimensional mixture of Gaussians. This resolves a central open question in the area, which has a history going back to a famous work of Pearson from 1894. 2) Beyond worst-case combinatorial optimization, where this method has led to new efficient algorithms that escape worst-case hardness while avoiding "overfitting" to brittle properties of any specific random model. Most recently, this line of work resulted in a resolution of longstanding open questions of finding optimal algorithms for "smoothed" models of k-SAT and "semirandom" models of Max-Clique. Taken together, these results suggest a unified theory for average-case algorithm design that not only makes substantial progress on long open foundational challenges but also brings a conceptual unity to algorithm design that we had never anticipated. Bio: Pravesh Kothari is an Assistant Professor of Computer Science at Carnegie Mellon University since September 2019. Before joining CMU, he was a postdoctoral Research Instructor jointly hosted by Princeton University and the Institute for Advanced Study from 2016-19. He obtained his Ph.D. in 2016 from the University of Texas at Austin. Kothari's recent work has focused on algorithm design for problems with statistical inputs. It is also the subject of his recent monograph "Semialgebraic Proofs and Efficient Algorithm Design". His research has been recognized with a Simons Award for graduate students in Theoretical Computer Science, a Google Research Scholar Award, an NSF CAREER Award, and an Alfred P. Sloan Research Fellowship.
Here are next week's CS Department Colloquium Series talks. As always, you can find the full calendar of events here: https://www.cs.princeton.edu/general/newsevents/events Speaker: Yao Lu, Microsoft Research Date: Monday, April 17 Time: 12:30pm EST Location: CS 105 Host: Kai Li Event page: https://www.cs.princeton.edu/events/26389 Title: Towards Intelligent Data Systems Abstract: From single-box databases, data systems are evolving into multi-tenant compute and storage platforms that host not only structured data analytics but also AI workloads and AI-enhanced system components. The result of this evolution, which I call an “intelligent” data system, creates new opportunities and challenges for research and production at the intersection of machine learning and systems. Key considerations in these systems include efficiency and cost, ML support and a flexible runtime for heterogeneous jobs. I will describe our work on query optimizers both for AI and aided by AI. For ML inference workloads over unstructured data, our optimizer injects proxy models for queries with complex predicates leading to a many-fold improvement in processing time; for query optimization in classic data analytics, our pre-trained models summarize structured datasets, answer cardinality estimation calls, and avoid the high training cost in recent instance-optimized database components. I will also describe our query processor and optimizer that enable and accelerate ML inference workflows on hybrid/IoT cloud. These efforts, combined with a few missing pieces that I will outline, contribute to better data systems where users can build, deploy, and optimize data analytics and AI applications with ease. Bio: Yao Lu is a researcher at the Data Systems group, Microsoft Research Redmond. He works at the intersection of machine learning and data systems towards improved data and compute platforms for cloud machine learning, as well as using machine learning to improve current data platforms. He received his Ph.D. from the University of Washington in 2018. Speaker: Ananya Kumar, Stanford University Date: Tuesday, April 18 Time: 12:30pm EST Location: CS 105 Host: Tom Griffiths Event page: https://www.cs.princeton.edu/events/26370 Title: Foundation Models for Robust Machine Learning Abstract: Machine learning systems are not robust—they suffer large drops in accuracy when deployed in different environments from what they were trained on. In this talk, I show that the foundation model paradigm—adapting models that are pretrained on broad unlabeled data—is a principled solution that leads to state-of-the-art robustness. I will focus on the key ingredients: how we should pretrain and adapt models for robustness. (1) First, I show that contrastive pretraining on unlabeled data learns transferable representations that improves accuracy even on domains where we had no labels. We explain why pretraining works in a very different way from some classical intuitions of collapsing representations (domain invariance). Our theory predicts phenomena on real datasets, and leads to improved pretraining methods. (1) Next, I will show that the standard approach of adaptation (updating all the model's parameters) can distort pretrained representations and perform poorly out-of-distribution. Our theoretical analysis leads to better methods for adaptation and state-of-the-art accuracies on ImageNet and in applications such as satellite remote sensing, wildlife conservation, and radiology. Bio: Ananya Kumar is a Ph.D. candidate in the Department of Computer Science at Stanford University, advised by Percy Liang and Tengyu Ma. His work focuses on representation learning, foundation models, and reliable machine learning. His papers have been recognized with several Spotlight and Oral presentations at NeurIPS, ICML, and ICLR, and his research is supported by a Stanford Graduate Fellowship. Speaker: Saadia Gabriel, University of Washington Date: Thursday, April 20 Time: 12:30pm EST Location: CS 105 Host: Olga Troyanskya Event page: https://www.cs.princeton.edu/events/26380 Title: Socially Responsible and Factual Reasoning for Equitable AI Systems Abstract: Understanding the implications underlying a text is critical to assessing its impact. This requires endowing artificial intelligence (AI) systems with pragmatic reasoning, for example to infer that the statement “Epidemics and cases of disease in the 21st century are “staged”” relates to unfounded conspiracy theories. In this talk, I discuss how shortcomings in the ability of current AI systems to reason about pragmatics leads to inequitable detection of false or harmful language. I demonstrate how these shortcomings can be addressed by imposing human-interpretable structure on deep learning architectures using insights from linguistics. In the first part of the talk, I describe how adversarial text generation algorithms can be used to improve model robustness. I then introduce a pragmatic formalism for reasoning about harmful implications conveyed by social media text. I show how this pragmatic approach can be combined with generative neural language models to uncover implications of news headlines. I also address the bottleneck to progress in text generation posed by gaps in evaluation of factuality. I conclude with an interdisciplinary study showing how content moderation informed by pragmatics can be used to ensure safe interactions with conversational agents, and my future vision for development of context-aware systems. Bio: Saadia Gabriel is a PhD candidate in the Paul G. Allen School of Computer Science & Engineering at the University of Washington, advised by Prof. Yejin Choi and Prof. Franziska Roesner. Her research revolves around natural language processing and machine learning, with a particular focus on building systems for understanding how social commonsense manifests in text (i.e. how do people typically behave in social scenarios), as well as mitigating spread of false or harmful text (e.g. Covid-19 misinformation). Her work has been covered by a wide range of media outlets like Forbes and TechCrunch. It has also received a 2019 ACL best short paper nomination, a 2019 IROS RoboCup best paper nomination and won a best paper award at the 2020 WeCNLP summit. Prior to her PhD, Saadia received a BA summa cum laude from Mount Holyoke College in Computer Science and Mathematics.
Here are next week's CS Department Colloquium Series talks. As always, you can find the full calendar of events here: https://www.cs.princeton.edu/general/newsevents/events CS Distinguished Colloquium Speaker Speaker: Corey Sanders '04, Microsoft Date: Monday, April 24 Time: 12:30pm EST Location: CS 105 Host: Adam Finkelstein Event page: https://www.cs.princeton.edu/events/26405 Title: The Future of Cloud Infrastructure for Large AI Models Abstract: Corey Sanders, CVP, Microsoft Cloud for Industry, will join us to share how Microsoft is building the infrastructure requirements for scaling AI and Large Language Model (LLM) services, with a specific focus on Azure, GPU sourcing and the architecture of OpenAI-specific data centers. Corey will highlight the impact of advanced AI models, such as Github CoPilot, including the workings and quality of these tools and models with real-world examples of how advanced AI models are already transforming the software development landscape. Bio: Corey Sanders is the Corporate Vice President for Microsoft Cloud for Industry, an organization dedicated to serving our customers with tailored industry solutions as they transform into successful digital businesses. Prior to this role, Corey led Microsoft Commercial Solution Areas, owning sales strategy and corporate technical sales across Solution Areas and Teams that include Digital Application Innovation, Azure Infrastructure & IoT, Azure Data & AI, Business Applications, Security, and Modern Workplace. His focus also included selling the full value of Microsoft cross-cloud solutions and advancing the technical depth of the Microsoft Solutions team. Earlier, Corey was Head of Product for Azure Compute and the founder of Microsoft Azure’s Infrastructure as a Service (IaaS) business. During that time, he was responsible for products, strategy and technical vision aligned to core Azure compute services. He also led program management for multiple Azure services. Earlier in his career, Corey was a developer in the Windows Serviceability team with ownership across the networking and kernel stack for Windows. In his first role at Microsoft in 2003, Corey served as an intern on the Windows team, after graduating from Princeton University, where he earned his Bachelor S.E. in Computer Science. Today, Corey resides with his family in New Jersey. CS Colloquium Speaker Speaker: Kaiming He, Facebook AI Research (FAIR) Date: Tuesday, April 25 Time: 12:30pm EST Location: CS 105 Host: Jia Deng / Kai Li Event page: https://www.cs.princeton.edu/events/26383 Title: In Pursuit of Visual Intelligence Abstract: Last decade's deep learning revolution in part began in the area of computer vision. The intrinsic complexity of visual perception problems urged the community to explore effective methods for learning abstractions from data. In this talk, I will review a few major breakthroughs that stemmed from computer vision. I will discuss my work on Deep Residual Networks (ResNets) that enabled deep learning to get way deeper, and its influence on the broader artificial intelligence areas over the years. I will also review my work on enabling deep learning to solve complex object detection and segmentation problems in simple and intuitive ways. On top of this progress, I will introduce recent research on learning from visual observations without human supervision, a topic known as visual self-supervised learning. I will discuss my research that contributed to shaping the two frontier directions on this topic. This research sheds light on future directions. I will discuss the opportunities for self-supervised learning in the visual world. I will also discuss how the research on computer vision may continue influencing broader areas, e.g., by generalizing self-supervised learning to scientific observations from nature. Bio: Kaiming He is a Research Scientist Director at Facebook AI Research (FAIR). Before joining FAIR in 2016, he was with Microsoft Research Asia from 2011 to 2016. He received his PhD degree from the Chinese University of Hong Kong in 2011, and his B.S. degree from Tsinghua University in 2007. His research areas include deep learning and computer vision. He is best-known for his work on Deep Residual Networks (ResNets), which have made significant impact on computer vision and broader artificial intelligence. He received several outstanding paper awards at top-tier conferences, including CVPR, ICCV, and ECCV. He received the PAMI Young Researcher Award in 2018. His publications have over 400,000 citations.
participants (1)
-
Emily C. Lawrence