Select a calendar:
Filter March Events by Event Type:
SUNMONTUEWEDTHUFRISAT
Events for the 4th week of March
-
Seminar will be exclusively online (no in-room presentation) - CS Colloquium: Simon S. Du (Princeton University) - Foundations of Learning Systems with (Deep) Function Approximators
Tue, Mar 24, 2020 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Simon S. Du, Princeton University
Talk Title: Foundations of Learning Systems with (Deep) Function Approximators
Series: CS Colloquium
Abstract: Function approximators, such as deep neural networks, play a crucial role in building learning systems that make predictions and decisions. In this talk, I will discuss my work on understanding, designing, and applying function approximators.
First, I will focus on understanding deep neural networks. The main result is that the over-parameterized neural network is equivalent to a new kernel, Neural Tangent Kernel. This equivalence implies two surprising phenomena: 1) the simple algorithm gradient descent provably finds the global optimum of the highly non-convex empirical risk, and 2) the learned neural network generalizes well despite being highly over-parameterized. Furthermore, this equivalence helps us design a new class of function approximators: we transform (fully-connected and graph) neural networks to (fully-connected and graph) Neural Tangent Kernels, which achieve superior performance on standard benchmarks.
In the second part of the talk, I will focus on applying function approximators to decision-making, aka reinforcement learning, problems. In sharp contrast to the (simpler) supervised prediction problems, solving reinforcement learning problems requires an exponential number of samples, even if one applies function approximators. I will then discuss what additional structures that permit statistically efficient algorithms.
This lecture satisfies requirements for CSCI 591: Research Colloquium
Biography: Simon S. Du is a postdoc at the Institute for Advanced Study of Princeton, hosted by Sanjeev Arora. He completed his Ph.D. in Machine Learning at Carnegie Mellon University, where he was co-advised by Aarti Singh and Barnabás Póczos. Previously, he studied EECS and EMS at UC Berkeley. He has also spent time at Simons Institute and research labs of Facebook, Google, and Microsoft. His research interests are broadly in machine learning, with a focus on the foundations of deep learning and reinforcement learning.
Host: Haipeng Luo
Location: Seminar will be exclusively online (no in-room presentation)
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
Computer Science General Faculty Meeting
Wed, Mar 25, 2020 @ 12:00 PM - 02:00 PM
Thomas Lord Department of Computer Science
Receptions & Special Events
Bi-Weekly regular faculty meeting for invited full-time Computer Science faculty only. Event details emailed directly to attendees.
Audiences: Invited Faculty Only
Contact: Assistant to CS chair
-
Seminar will be exclusively online (no in-room presentation) - CS Colloquium: Farnaz Behrang (Georgia Institute of Technology) - Leveraging Existing Software Artifacts to Support Design, Development, and Testing of Mobile Applications
Wed, Mar 25, 2020 @ 02:00 PM - 03:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Farnaz Behrang, Georgia Institute of Technology
Talk Title: Leveraging Existing Software Artifacts to Support Design, Development, and Testing of Mobile Applications
Series: CS Colloquium
Abstract: We are living in the era of big data, in which generating and sharing data has become much easier, and massive amounts of information are created in a fraction of a second. In the context of software engineering, in particular, the number of open-source software repositories (e.g., GitHub, Bitbucket, SourceForge) where software developers share their software artifacts is ever-increasing, and hundreds of millions of lines of code are freely available and easily accessible. This has resulted in an increasing interest in analyzing the rich data available in such repositories. In the past decade, researchers have been mining online repositories to take advantage of existing source code to support different development activities, such as bug prediction, refactoring, and API updates. Despite the large number of proposed techniques that leverage existing source code, however, these techniques mostly focus on supporting coding activities. Other important software engineering tasks, such as software design and testing, have been mostly ignored by previous work.
In this talk, I will present my research on leveraging existing source code and other related artifacts (e.g., test cases) to support the design, development, and testing of mobile applications using automated techniques. I will first present a technique that leverages the growing number of open-source apps in public repositories to support app design and development. I will then present techniques that take advantage of existing test cases to reduce the cost of testing mobile apps. I will conclude my talk sketching future research directions that I plan to pursue.
This lecture satisfies requirements for CSCI 591: Research Colloquium
Biography: Farnaz Behrang is a Ph.D. candidate in the School of Computer Science at the Georgia Institute of Technology. Her research interests lie primarily in the area of software engineering, with a focus on software analysis and testing. Her research goal is to develop automated techniques and tools that improve software quality and developer productivity. Her work has been recognized with several awards including ACM SIGSOFT Distinguished Paper Awards at MOBILESOFT 2018 and FSE 2015.
Host: Chao Wang
Location: Seminar will be exclusively online (no in-room presentation)
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
Seminar will be exclusively online (no in-room presentation) - CS Colloquium: Aditya Grover (Stanford University) - Machine Learning for Accelerating Scientific Discovery
Thu, Mar 26, 2020 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Aditya Grover, Stanford University
Talk Title: Machine Learning for Accelerating Scientific Discovery
Series: CS Colloquium
Abstract: The dramatic increase in both sensor capabilities and computational power over the last few decades has created enormous opportunities for using machine learning (ML) to enhance scientific discovery. To realize this potential, ML systems must seamlessly integrate with the key tools for scientific discovery. For instance, how can we incorporate scientific domain knowledge within ML algorithms? How can we use ML to quantify uncertainty in simulations? How can we use ML to plan experiments under real-world budget constraints? For these questions, I'll first present the key computational and statistical challenges through the lens of probabilistic modeling. Next, I'll highlight limitations of existing approaches for scaling to high-dimensional data and present algorithms from my research that can effectively overcome these challenges. These algorithms are theoretically principled, domain-agnostic, and exhibit strong empirical performance. Notably, I'll describe a collaboration with chemists and material scientists where we used probabilistic models to efficiently optimize an experimental pipeline for electric batteries. Finally, I'll conclude with an overview of future opportunities for using ML to accelerate scientific discovery.
This lecture satisfies requirements for CSCI 591: Research Colloquium
Biography: Aditya Grover is a fifth-year Ph.D. candidate in Computer Science at Stanford University advised by Stefano Ermon. His research focuses on probabilistic modeling and reasoning and is grounded in real-world scientific applications. Aditya's research has been published in top scientific and ML/AI venues (e.g., Nature, NeurIPS, ICML, ICLR, AAAI, AISTATS), included in widely-used open source ML software, and deployed into production at major technology companies. His work has been recognized with a best paper award (StarAI), a Lieberman Fellowship, a Data Science Institute Scholarship, and a Microsoft Research Ph.D. Fellowship. He is also a Teaching Fellow at Stanford since 2018, where he co-created and teaches a new class on Deep Generative Models. Previously, Aditya obtained his bachelors in Computer Science and Engineering from IIT Delhi in 2015, where he received a best undergraduate thesis award.
Host: Bistra Dilkina
Location: Seminar will be exclusively online (no in-room presentation)
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
Seminar will be exclusively online (no in-room presentation) - CS Colloquium: Zhihao Jia (Stanford University) - Automated Discovery of Machine Learning Optimizations
Thu, Mar 26, 2020 @ 04:00 PM - 05:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Zhihao Jia, Stanford University
Talk Title: Automated Discovery of Machine Learning Optimizations
Series: CS Colloquium
Abstract: As an increasingly important workload, machine learning (ML) applications require different performance optimization techniques from traditional runtimes and compilers. In particular, to accelerate ML applications, it is generally necessary to perform ML computations on heterogeneous hardware and parallelize computations using multiple data dimensions, neither of which is even expressible in traditional compilers and runtimes. In this talk, I will describe my work on automated discovery of performance optimizations to accelerate ML computations.
TASO, the Tensor Algebra SuperOptimizer, optimizes the computation graphs of deep neural networks (DNNs) by automatically generating potential graph optimizations and formally verifying their correctness. TASO outperforms rule-based graph optimizers in existing ML systems (e.g., TensorFlow, TensorRT, and TVM) by up to 3x by automatically discovering novel graph optimizations, while also requiring significantly less human effort.
FlexFlow is a system for accelerating distributed DNN training. FlexFlow identifies parallelization dimensions not considered in existing ML systems (e.g., TensorFlow and PyTorch) and automatically discovers fast parallelization strategies for a specific parallel machine. Companies and national labs are using FlexFlow to train production ML models that do not scale well in current ML systems, achieving over 10x performance improvement.
I will also outline future research directions for further automating ML systems, such as codesigning ML models, software systems, and hardware backends for end-to-end ML deployment.
This lecture satisfies requirements for CSCI 591: Research Colloquium.
Biography: Zhihao Jia is a Ph.D. candidate in the Computer Science department at Stanford University working with Alex Aiken and Matei Zaharia. His research interests lie in the intersection of computer systems and machine learning, with a focus on building efficient, scalable, and high-performance systems for ML computations.
Host: Leana Golubchik
Location: Seminar will be exclusively online (no in-room presentation)
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
