- talks - lists.cs.princeton.edu

Colloquium Speaker-Chris Maddison, Tuesday, April 9 at 12:30pm
by Emily C. Lawrence 03 Apr '19

03 Apr '19

Colloquium Speaker Chris Maddison from University of Oxford Tuesday, April 9 - 12:30pm Computer Science - Room 105 Host: Ryan Adams AlphaGo and the Computational Challenges of Machine Learning [ http://www.cs.princeton.edu/events/25841 | http://www.cs.princeton.edu/events/25841 ] Many computational challenges in machine learning involve the three problems of optimization, integration, and fixed-point computation. These three can often be reduced to each other, so they may also provide distinct vantages on a single problem. In this talk, I present a small part of this picture through a discussion of my work on AlphaGo and two vignettes on my work on the interplay between optimization and Monte Carlo. AlphaGo is the first computer program to defeat a world-champion player, Lee Sedol , in the board game of Go. My work laid the groundwork of the neural net components of AlphaGo , and culminated in our Nature publication describing AlphaGo's algorithm, at whose core hide these three problems. In the first vignette, I present the Hamiltonian descent methods we introduced for first-order optimization. These methods are inspired by the Monte Carlo literature and can achieve fast linear convergence without strong convexity by using a non-standard kinetic energy to condition the optimization. In the second vignette I cover our A* Sampling method, which reduces the problem of Monte Carlo simulation to an optimization problem, and an application to gradient estimation in stochastic computation graphs. Bio: Chris Maddison is a PhD candidate in the Statistical Machine Learning Group in the Department of Statistics at the University of Oxford. He is an Open Philanthropy AI Fellow and spends two days a week as a Research Scientist at DeepMind. His research is broadly focused on the development of numerical methods for deep learning and machine learning. He has worked on methods for variational inference, numerical optimization, and Monte Carlo estimation with a specific focus on those that might work at scale with few assumptions. Chris received his MSc. from the University of Toronto. He received a NeurIPS Best Paper Award in 2014, and was one of the founding members of the AlphaGo project.

1 0

REMINDER: Colloquium Speaker-Corey Sanders '04, tomorrow at 4:30pm
by Emily C. Lawrence 02 Apr '19

02 Apr '19

Distinguished Colloquium Speaker Corey Sanders '04 from Microsoft Solutions Wednesday, April 3 - 4:30pm Computer Science - Room 104 (large auditorium) AI and Reshaping the Industry [ http://www.cs.princeton.edu/events/25839 | http://www.cs.princeton.edu/events/25839 ] Artificial Intelligence has quickly moved from being a thing of science fiction to driving business forward across a wide variety of industries including retail, manufacturing, oil and gas, health, and financial services. It is becoming a ubiquitous component of normal customer IT, whether to improve end-customer experience or reduce back-end costs. In this talk, Corey will walk through some real and technical examples of AI and how it is reshaping every industry. Corey will also delve into the responsibility we all have in considering the impact of this technology as it continues to advance and improve. Bio: Corey is the Corporate Vice President for Microsoft Solutions. He is responsible for the sales strategy and corporate technical sales team for Azure, M365, and Dynamics. Previously, Corey was the Head of Product for Azure Compute, responsible for product, strategy, and technical vision for core Azure compute services Azure Virtual Machines (both Windows and Linux), Containers, Kubernetes, OSS, Service Fabric, Event Grid, Service Bus, Management, Migration and serverless computing such as Functions. Since joining Microsoft in 2004 Corey has led program management for multiple Azure services, was a developer in Windows Serviceability team owning areas across the networking and kernel stack for Windows and was the founder of Azure’s Infrastructure-as-a -service business. Corey graduated from Princeton in 2004 with a BSE in Computer Science.

1 0

Colloquium Speaker-Jiajun Wu, TODAY at 12:30pm
by Emily C. Lawrence 01 Apr '19

01 Apr '19

Colloquium Speaker Jiajun Wu from the Massachusetts Institute of Technology Monday, April 1 - 12:30pm Computer Science - Room 105 Host: Tom Griffiths Learning to See the Physical World [ http://www.cs.princeton.edu/events/25840 | http://www.cs.princeton.edu/events/25840 ] Human intelligence is beyond pattern recognition. From a single image, we're able to explain what we see, reconstruct the scene in 3D, predict what's going to happen, and plan our actions accordingly. In this talk, I will present our recent work on physical scene understanding---building versatile, data-efficient, and generalizable machines that learn to see, reason about, and interact with the physical world. The core idea is to exploit the generic, causal structure behind the world, including knowledge from computer graphics, physics, and language, in the form of approximate simulation engines, and to integrate them with deep learning. Here, deep learning plays two major roles: first, it learns to invert simulation engines for efficient inference; second, it learns to augment simulation engines for constructing powerful forward models. I'll focus on a few topics to demonstrate this idea: building scene representation for both object geometry and physics; learning expressive dynamics models for planning and control; perception and reasoning beyond vision. Bio: Jiajun Wu is a Ph.D. student in Electrical Engineering and Computer Science at Massachusetts Institute of Technology. He received his B.Eng. from Tsinghua University in 2014. His research interests lie in the intersection of computer vision, machine learning, robotics, and computational cognitive science. His research has been recognized through the IROS Best Paper Award on Cognitive Robotics and fellowships from Facebook, Nvidia , Samsung, Baidu, and Adobe, and his work has been covered by major media outlets including CNN, BBC, WIRED, and MIT Tech Review.

1 0

Colloquium Speaker-David van Dijk, Friday, April 5 at 12:30pm
by Emily C. Lawrence 01 Apr '19

01 Apr '19

David van Dijk from Yale University Friday, April 5 - 12:30pm Computer Science - Room 105 [ https://www.cs.princeton.edu/events/25845 | https://www.cs.princeton.edu/events/25845 ] Manifold learning uncovers hidden structure in complex cellular state space In the era of big biological data, there is a pressing need for methods that visualize, integrate and interpret high-throughput high-dimensional data to enable biological discovery. There are several major challenges in analyzing high-throughput biological data. These include the curse of (high) dimensionality, noise, sparsity, missing values, bias, and collection artifacts. In my work, I try to solve these problems using computational methods that are based on manifold learning. A manifold is a smoothly varying low-dimensional structure embedded within high-dimensional ambient measurement space. In my talk, I will present a number of my recently completed and ongoing projects that utilize the manifold, implemented using graph signal processing and deep learning, to understand large biomedical datasets. These include MAGIC, a data denoising and imputation method designed to ‘fix’ single-cell RNA-sequencing data, PHATE, a dimensionality reduction and visualization method specifically designed to reveal continuous progression structure, and two deep learning methods that use specially designed constraints to allow for deep interpretable representations of heterogeneous systems. I will demonstrate that these methods can give insight into diverse biological systems such as breast cancer epithelial-to-mesenchymal transition, human embryonic stem cell development, the gut microbiome, and tumor infiltrating lymphocytes.

1 0

REMINDER: Colloquium Speaker-Jiajun Wu, tomorrow at 12:30pm
by Emily C. Lawrence 31 Mar '19

31 Mar '19

Colloquium Speaker Jiajun Wu from the Massachusetts Institute of Technology Monday, April 1 - 12:30pm Computer Science - Room 105 Host: Tom Griffiths Learning to See the Physical World [ http://www.cs.princeton.edu/events/25840 | http://www.cs.princeton.edu/events/25840 ] Human intelligence is beyond pattern recognition. From a single image, we're able to explain what we see, reconstruct the scene in 3D, predict what's going to happen, and plan our actions accordingly. In this talk, I will present our recent work on physical scene understanding---building versatile, data-efficient, and generalizable machines that learn to see, reason about, and interact with the physical world. The core idea is to exploit the generic, causal structure behind the world, including knowledge from computer graphics, physics, and language, in the form of approximate simulation engines, and to integrate them with deep learning. Here, deep learning plays two major roles: first, it learns to invert simulation engines for efficient inference; second, it learns to augment simulation engines for constructing powerful forward models. I'll focus on a few topics to demonstrate this idea: building scene representation for both object geometry and physics; learning expressive dynamics models for planning and control; perception and reasoning beyond vision. Bio: Jiajun Wu is a Ph.D. student in Electrical Engineering and Computer Science at Massachusetts Institute of Technology. He received his B.Eng. from Tsinghua University in 2014. His research interests lie in the intersection of computer vision, machine learning, robotics, and computational cognitive science. His research has been recognized through the IROS Best Paper Award on Cognitive Robotics and fellowships from Facebook, Nvidia , Samsung, Baidu, and Adobe, and his work has been covered by major media outlets including CNN, BBC, WIRED, and MIT Tech Review.

1 0

Colloquium Speaker-Henry Corrigan-Gibbs, Thursday, April 4 at 12:30pm
by Emily C. Lawrence 29 Mar '19

29 Mar '19

Colloquium Speaker Henry Corrigan-Gibbs from Stanford University Thursday, April 4 - 12:30pm Computer Science - Room 105 Host: Arvind Narayanan [ https://www.cs.princeton.edu/events/25830 | https://www.cs.princeton.edu/events/25830 ] Protecting Privacy by Splitting Trust When the maker of my phone, smart-watch, or web browser collects data about how I use it, must I trust the manufacturer to protect that sensitive information from theft? When I use the cryptographic hardware module in my laptop, need I trust that it will keep my secrets safe? When I use a messaging app to chat with friends, must I trust the app vendor not to sell the details of my messaging activity for profit? This talk will show that we can get the functionality we want from our systems without having to put blind faith in the correct behavior of the companies collecting our data, building our hardware, or designing our apps. The principle is to split our trust -- among organizations, or devices, or users. I will introduce new cryptographic techniques and systems-level optimizations that make it practical to split trust in a variety of settings. Then, I will present three built systems that employ these ideas, including one that now ships with the Firefox browser. Bio: Henry Corrigan-Gibbs is a Ph.D. candidate at Stanford, advised by Dan Boneh . His research interests are in computer security, applied cryptography, and online privacy. Henry and his collaborators have received the Best Young Researcher Paper Award at Eurocrypt 2018, the 2016 Caspar Bowden Award for Outstanding Research in Privacy Enhancing Technologies, and the 2015 IEEE Security and Privacy Distinguished Paper Award, and Henry's work has been cited by IETF and NIST.

1 0

Distinguished Colloquium Speaker-Corey Sanders '04, Wednesday, April 3 at 4:30pm
by Emily C. Lawrence 28 Mar '19

28 Mar '19

Distinguished Colloquium Speaker Corey Sanders '04 from Microsoft Solutions Wednesday, April 3 - 4:30pm Computer Science - Room 104 (large auditorium) AI and Reshaping the Industry [ http://www.cs.princeton.edu/events/25839 | http://www.cs.princeton.edu/events/25839 ] Artificial Intelligence has quickly moved from being a thing of science fiction to driving business forward across a wide variety of industries including retail, manufacturing, oil and gas, health, and financial services. It is becoming a ubiquitous component of normal customer IT, whether to improve end-customer experience or reduce back-end costs. In this talk, Corey will walk through some real and technical examples of AI and how it is reshaping every industry. Corey will also delve into the responsibility we all have in considering the impact of this technology as it continues to advance and improve. Bio: Corey is the Corporate Vice President for Microsoft Solutions. He is responsible for the sales strategy and corporate technical sales team for Azure, M365, and Dynamics. Previously, Corey was the Head of Product for Azure Compute, responsible for product, strategy, and technical vision for core Azure compute services Azure Virtual Machines (both Windows and Linux), Containers, Kubernetes, OSS, Service Fabric, Event Grid, Service Bus, Management, Migration and serverless computing such as Functions. Since joining Microsoft in 2004 Corey has led program management for multiple Azure services, was a developer in Windows Serviceability team owning areas across the networking and kernel stack for Windows and was the founder of Azure’s Infrastructure-as-a -service business. Corey graduated from Princeton in 2004 with a BSE in Computer Science.

1 0

TODAY March 27 at 4:30pm in B205 E-Quad - Min Ye, Princeton University
by Emily Lawrence 27 Mar '19

27 Mar '19

EE SEMINAR SERIES Speaker: Min Ye, Princeton University Title: Long-standing problems and new challenges in coding theory Day: Wednesday, March 27, 2019 Time: 4:30 pm Room: B205 Engineering Quadrangle Host: Prof. Peter Ramadge Abstract: Emerging problems in data analysis and management create a demand for new methods of data storage, aggregation, and transmission. Systems as diverse as data centers or wireless networks have put forward new requirements to the encoding and representation of digital information. In this talk, we discuss three problems arising from this modern coding-theoretic context: (i) We present coding schemes that minimize communication costs in distributed storage and learning systems. Our schemes were evaluated on Amazon EC2 clusters and were shown to improve upon the state-of-the-art solutions. (ii) We address long-standing coding-theoretic problems regarding Reed-Muller codes: We present advances on the capacity-achieving conjecture as well as a new decoder of Reed-Muller codes which significantly outperforms the state-of-the-art codes adopted in the 5G standard. (iii) We discuss a new method of using neural networks with certain permutation-invariant properties to decode codes with symmetric structures, which suggests a way to construct and efficiently decode the codes with the best possible finite length performance. Bio: Min Ye received his B.S. in Electrical Engineering from Peking University, Beijing, China in 2012, and his Ph.D. in the Department of Electrical and Computer Engineering, University of Maryland, College Park in 2017. He is currently a postdoctoral researcher at Princeton University. His research interests include coding theory, information theory, differential privacy, and machine learning.

1 0

Yichen Chen will present his Pre FPO on Monday, April 1, 2019 at 2pm in CS 105.
by Nicki Mahler 26 Mar '19

26 Mar '19

Yichen Chen will present his Pre FPO on Monday, April 1, 2019 at 2pm in CS 105. The members of his committee are as follows: Mengdi Wang (advisor), Sanjeev Arora (nonreader), Robert Schapire (nonreader), Karthik Narasimhan (reader), Yuxin Chen (reader) The talk title and abstract follow below. All are welcome to attend. Title: Sample-efficient reinforcement learning via utilizing state features Abstract: The field of reinforcement learning has made enormous progress in the last few years, largely due to deep neural networks. Despite success on many practical problems, our reinforcement learning algorithms require huge amount of data. In many applications such as healthcare, data of this large scale is simply not available. In this talk, I explore novel approaches from stochastic optimization that open up opportunities for designing sample-efficient reinforcement learning algorithms. First, I will show how state features, which might come from machine learning algorithms or expert domain knowledge, can facilitate the design of data-efficient reinforcement learning algorithms. Next, I will show that reinforcement learning problems are easier to solve compared to Markov decision processes with full knowledge due to the powerful sampling mechanism in reinforcement learning.

1 0

Colloquium Speaker-Jiajun Wu, Monday, April 1 at 12:30pm
by Emily C. Lawrence 26 Mar '19

26 Mar '19

Colloquium Speaker Jiajun Wu from the Massachusetts Institute of Technology Monday, April 1 - 12:30pm Computer Science - Room 105 Host: Tom Griffiths Learning to See the Physical World [ http://www.cs.princeton.edu/events/25840 | http://www.cs.princeton.edu/events/25840 ] Human intelligence is beyond pattern recognition. From a single image, we're able to explain what we see, reconstruct the scene in 3D, predict what's going to happen, and plan our actions accordingly. In this talk, I will present our recent work on physical scene understanding---building versatile, data-efficient, and generalizable machines that learn to see, reason about, and interact with the physical world. The core idea is to exploit the generic, causal structure behind the world, including knowledge from computer graphics, physics, and language, in the form of approximate simulation engines, and to integrate them with deep learning. Here, deep learning plays two major roles: first, it learns to invert simulation engines for efficient inference; second, it learns to augment simulation engines for constructing powerful forward models. I'll focus on a few topics to demonstrate this idea: building scene representation for both object geometry and physics; learning expressive dynamics models for planning and control; perception and reasoning beyond vision. Bio: Jiajun Wu is a Ph.D. student in Electrical Engineering and Computer Science at Massachusetts Institute of Technology. He received his B.Eng. from Tsinghua University in 2014. His research interests lie in the intersection of computer vision, machine learning, robotics, and computational cognitive science. His research has been recognized through the IROS Best Paper Award on Cognitive Robotics and fellowships from Facebook, Nvidia , Samsung, Baidu, and Adobe, and his work has been covered by major media outlets including CNN, BBC, WIRED, and MIT Tech Review.

1 0