- talks - lists.cs.princeton.edu

Colloquium Speaker Ramón Cáceres TODAY at 1:30pm
by Emily C. Lawrence 16 Oct '19

16 Oct '19

Distinguished CS Colloquium Speaker Ramón Cáceres from Google Wednesday, October 16 - 1:30pm Computer Science - Room 105 Host: Margaret Martonosi [ https://www.cs.princeton.edu/events/25878 | https://www.cs.princeton.edu/events/25878 ] Zanzibar: Google’s Consistent, Global Authorization System Determining whether online users are authorized to access digital objects is central to preserving privacy. This talk presents the design, implementation, and deployment of Zanzibar, a global system for storing and evaluating access control lists. Zanzibar provides a uniform data model and configuration language for expressing a wide range of access control policies from hundreds of client services at Google, including Calendar, Cloud, Drive, Maps, Photos, and YouTube. Its authorization decisions respect causal ordering of user actions and thus provide external consistency amid changes to access control lists and object contents. Zanzibar scales to trillions of access control lists and millions of authorization requests per second to support services used by billions of people. It has maintained 95th-percentile latency of less than 10 milliseconds and availability of greater than 99.999% over 3 years of production use. Bio: Ramón Cáceres is a software engineer at Google. He was previously a researcher at AT&T Labs and IBM Research. His areas of focus include computer systems and networks, mobile computing services and applications, and location data analysis and modeling. He holds a PhD in Computer Science from the University of California at Berkeley. He is an IEEE Fellow and an ACM Distinguished Scientist.

1 0

Colloquium Speaker Dougal Maclaurin TODAY at 12:30
by Emily C. Lawrence 15 Oct '19

15 Oct '19

CS Colloquium Speaker Dougal Maclaurin from Google Brain Tuesday, October 15 - 12:30pm Computer Science - Room 105 Host: Ryan Adams [ https://www.cs.princeton.edu/events/25877 | https://www.cs.princeton.edu/events/25877 ] JAX: Accelerated machine learning research via composable function transformations in Python JAX is a system for high-performance machine learning research. It offers the familiarity of Python+NumPy and the speed of hardware accelerators, and it enables the definition and the composition of function transformations useful for machine learning programs. In particular, these transformations include automatic differentiation, automatic batching, end-to-end-compilation (via XLA), and parallelizing over multiple accelerators. They are the key to JAX's power and to its relative simplicity. JAX had its initial open-source release in December 2018 (https://github.com/google/jax). It is currently being used by several groups of researchers for a wide range of advanced applications, from studying spectra of neural networks, to probabilistic programming and Monte Carlo methods, and scientific applications in physics and biology. Users appreciate JAX most of all for its ease of use and flexibility. Bio: Dougal Maclaurin is a research scientist at Google. He works on programming languages and systems for machine learning, particularly the Python library JAX. He started Autograd , a system for automatic differentiation in Python, which has inspired the design of several systems, including PyTorch , MinPy , Torch Autograd and Julia Autograd . He is a co-founder of Day Zero Diagnostics, a startup developing a sequencing-based diagnostic for drug-resistant infections. He received his Ph.D. from Harvard in 2016, working with Ryan Adams on the development of methods for machine learning. His work on scalable MCMC, "Firefly Monte Carlo", was recognized with the Best Paper award at UAI 2014.

1 0

CS Colloquium Speaker Andrés Monroy-Hernández, Monday, October 21 at 12:30pm
by Emily C. Lawrence 15 Oct '19

15 Oct '19

CS Colloquium Speaker Andrés Monroy-Hernández from Snap Inc. Monday, October 21 - 12:30pm Computer Science - Room 105 Host: Adam Finkelstein [ https://www.cs.princeton.edu/events/25890 | https://www.cs.princeton.edu/events/25890 ] Designing and deploying social computing systems inside and outside the lab Social computing has permeated most aspects of our lives, from work to play. In recent years, however, social platforms have faced challenges ranging from labor concerns about crowd work to misinformation on social media. These and other challenges, along with the emergence of new technical platforms, create opportunities to re-imagine the future of the field. In this talk, I will discuss my research designing and deploying social computing systems to help people connect and collaborate to learn new skills, crowdsource news reporting, and delegate tasks to hybrid AI systems. First, I describe an online community I created to connect millions of young creators to learn computational thinking skills as they create, share, and remix games and animations on the web. Then I shift to how this work inspired my next set of systems to connect urban residents as they share news about their local communities. Finally, I will discuss a novel workplace productivity tool we created to enable information workers to delegate tasks to hybrid intelligence systems powered by humans and AI. I close by articulating the challenges and opportunities ahead for the field, and the ways my team is beginning to explore new systems that support more authentic and intimate connections, building on new technologies including wearables sensors and AR. Bio: Andrés Monroy-Hernández is a lead research scientist at Snap Inc , where he manages the human-computing interaction research team. He is also an affiliate professor at the University of Washington. His work focuses on the design and study of social computing technologies that help people connect and collaborate. Previously, he was at Microsoft Research for six years, where he led the research and development of Calendar.help , Microsoft’s first hybrid AI product. His research has received best paper awards at CHI, CSCW, HCOMP, and ICWSM, and has been featured in The New York Times, CNN, Wired, BBC, and The Economist. Andrés was named one of the 35 Innovators under 35 by the MIT Technology Review magazine in Latin America, and one the most influential Latinos in Tech by CNET. Andrés holds master’s and Ph.D. from MIT, where he led the creation of the Scratch online community. He has a BS from Tec de Monterrey in México. More info at [ http://andresmh.com/ | http://andresmh.com ]

1 0

Reminder: Colloquium Speaker Dougal Maclaurin tomorrow at 12:30
by Emily C. Lawrence 14 Oct '19

14 Oct '19

CS Colloquium Speaker Dougal Maclaurin from Google Brain Tuesday, October 15 - 12:30pm Computer Science - Room 105 Host: Ryan Adams [ https://www.cs.princeton.edu/events/25877 | https://www.cs.princeton.edu/events/25877 ] JAX: Accelerated machine learning research via composable function transformations in Python JAX is a system for high-performance machine learning research. It offers the familiarity of Python+NumPy and the speed of hardware accelerators, and it enables the definition and the composition of function transformations useful for machine learning programs. In particular, these transformations include automatic differentiation, automatic batching, end-to-end-compilation (via XLA), and parallelizing over multiple accelerators. They are the key to JAX's power and to its relative simplicity. JAX had its initial open-source release in December 2018 ( [ https://github.com/google/jax | https://github.com/google/jax ] ). It is currently being used by several groups of researchers for a wide range of advanced applications, from studying spectra of neural networks, to probabilistic programming and Monte Carlo methods, and scientific applications in physics and biology. Users appreciate JAX most of all for its ease of use and flexibility. Bio: Dougal Maclaurin is a research scientist at Google. He works on programming languages and systems for machine learning, particularly the Python library JAX. He started Autograd , a system for automatic differentiation in Python, which has inspired the design of several systems, including PyTorch , MinPy , Torch Autograd and Julia Autograd . He is a co-founder of Day Zero Diagnostics, a startup developing a sequencing-based diagnostic for drug-resistant infections. He received his Ph.D. from Harvard in 2016, working with Ryan Adams on the development of methods for machine learning. His work on scalable MCMC, "Firefly Monte Carlo", was recognized with the Best Paper award at UAI 2014.

1 0

CS Colloquium Speaker: Alexa Meade TODAY at 4:30pm
by Emily C. Lawrence 14 Oct '19

14 Oct '19

CS Colloquium Speaker Alexa Meade Monday, October 14 - 4:30pm James Stewart Theater Lewis Center for the Arts, 185 Nassau Street Hosts: Adam Finkelstein & Jeff Whetstone [ https://www.cs.princeton.edu/events/25887 | https://www.cs.princeton.edu/events/25887 ] Painting in More Dimensions Artist Alexa Meade paints on the human body and three-dimensional spaces, creating the illusion that our reality is a two-dimensional painting. She will discuss her artistic process and collaborations with mathematicians, magicians, and theoretical physicists. Her groundbreaking work has been exhibited around the world at the Grand Palais in Paris, the United Nations in New York, the Smithsonian National Portrait Gallery in Washington, DC and Shibuya Crossing in Tokyo. Her solo show on Rodeo Drive in Beverly Hills was attended by forty-thousand people. She has created large scale interactive installations at Coachella, Cannes Lions, and Art Basel. Alexa has been commissioned by BMW, Sony, Adidas, and the San Francisco Symphony Orchestra. She painted pop star Ariana Grande for her iconic “God is a Woman” music video, which has about 250 million views. Alexa has lectured at TED, Apple, and Stanford and accepted an invitation to the White House under President Obama. She has been honored with the "Disruptive Innovation Award" by the Tribeca Film Festival and has been Artist-in-Residence at both Google and the Perimeter Institute for Theoretical Physics. InStyle has named Alexa among their "Badass Women."

1 0

Reminder: CS Colloquium Speaker Alexa Meade, tomorrow at 4:30pm
by Emily C. Lawrence 13 Oct '19

13 Oct '19

CS Colloquium Speaker Alexa Meade Monday, October 14 - 4:30pm James Stewart Theater Lewis Center for the Arts, 185 Nassau Street Hosts: Adam Finkelstein & Jeff Whetstone [ https://www.cs.princeton.edu/events/25887 | https://www.cs.princeton.edu/events/25887 ] Painting in More Dimensions Artist Alexa Meade paints on the human body and three-dimensional spaces, creating the illusion that our reality is a two-dimensional painting. She will discuss her artistic process and collaborations with mathematicians, magicians, and theoretical physicists. Her groundbreaking work has been exhibited around the world at the Grand Palais in Paris, the United Nations in New York, the Smithsonian National Portrait Gallery in Washington, DC and Shibuya Crossing in Tokyo. Her solo show on Rodeo Drive in Beverly Hills was attended by forty-thousand people. She has created large scale interactive installations at Coachella, Cannes Lions, and Art Basel. Alexa has been commissioned by BMW, Sony, Adidas, and the San Francisco Symphony Orchestra. She painted pop star Ariana Grande for her iconic “God is a Woman” music video, which has about 250 million views. Alexa has lectured at TED, Apple, and Stanford and accepted an invitation to the White House under President Obama. She has been honored with the "Disruptive Innovation Award" by the Tribeca Film Festival and has been Artist-in-Residence at both Google and the Perimeter Institute for Theoretical Physics. InStyle has named Alexa among their "Badass Women."

1 0

Andy Zeng will present his FPO "Learning Visual Affordances for Robotic Manipulation" on Thursday, 10/24/2019 at 1:30pm in CS 402
by Nicki Mahler 10 Oct '19

10 Oct '19

Andy Zeng will present his FPO "Learning Visual Affordances for Robotic Manipulation" on Thursday, 10/24/2019 at 1:30pm in CS 402. The members of his committee are as follows: Thomas Funkhouser (Adviser); Examiners: Olga Russakovsky, Adam Finkelstein, and Thomas Funkhouser; Readers: Szymon Rusinkiewicz and Alberto Rodriguez (MIT). A copy of his thesis is available upon request. Please email ngotsis(a)cs.princeton if you would like a copy of the thesis. Everyone is invited to attend his talk. The talk title and abstract follow below. Title: Learning Visual Affordances for Robotic Manipulation Abstract: A humans remarkable ability to manipulate unfamiliar objects with little prior knowledge of them is a constant inspiration for robotics research. Despite the interest of the research community, and despite its practical value, robust manipulation of novel objects in cluttered environments still remains a largely unsolved problem. Classic solutions (e.g. involving 6D object pose estimation) typically require prior knowledge of the objects (e.g. class categories or 3D CAD models), which may not be available outside of highly constrained settings. More recent deep learning methods using end-to-end convolutional networks (e.g. raw pixels to motor torques) have the potential to model complex skills that generalize, but they remain highly data inefficient – and robot data (e.g. trial and error) is expensive. In this thesis, we consider an approach to learning manipulation called visual affordances. The idea is to use classic controllers to design motion primitives, then use convolutional networks to map from visual observations (e.g. images) to the perceived affordances (e.g. confidence scores or action-values) of the primitives for every pixel of the input. By leveraging dense equivariant state and action representations, this formulation can be used to acquire complex vision-based manipulation skills (e.g. pushing, grasping, throwing) on real robot platforms that generalize to novel objects, while using orders of magnitude less data. While visual affordances may not be directly compatible with classic planning frameworks that involve explicit forward simulation or propagation, in this thesis we show that it is possible to workaround this limitation by extending it with model-free reinforcement learning to sequence primitive picking motions for more complex manipulation policies. We also study how it can be combined with residual physics (learning to predict residual values on top of control parameter estimates from an initial analytical controller) to enable learning end-to-end visuomotor policies that leverage the benefits of analytical models while still maintaining the capacity (via data-driven residuals) to account for real-world dynamics that are not explicitly modeled. Finally, we conclude by discussing the limitations of learning visual affordances, which suggest directions for future work.

1 0

Zoe Ashwood will present her General Exam on Thursday, October 17, 2019 at 10am in PNI 159
by Nicki Mahler 10 Oct '19

10 Oct '19

Zoe Ashwood will present her General Exam on Thursday, October 17, 2019 at 10am in PNI 159. The members of her committee are as follows: Jonathan Pillow (adviser), Sebastian Seung, and Tom Griffiths Everyone is invited to attend her talk, and those faculty wishing to remain for the oral exam following are welcome to do so. Her abstract and reading list follow below. Title: Characterizing decision-making with Hidden Markov Models and Generalized Linear Models Abstract: Think about the number of decisions you made this morning, even before leaving the house. You had to decide what to wear, what to eat for breakfast, and whether to bike to work or take the bus. Why do we make the decisions that we do? How do things like past experience and internal state (whether we feel stressed or elated, for example) influence our choices? What if the options for the decision are perturbed slightly, and instead of having to choose between our favorite breakfast cereal and toast, the choice is now between the supermarket’s own-brand version of this cereal and toast? Would we still disregard the toast? In my talk, I will describe our approach to modeling the choice behavior of mice and humans in two laboratory perceptual decision-making tasks. Whereas existing models of choice behavior (Busse et al., 2011; Akrami et al., 2018) assume that an animal is either engaged in the task or guessing randomly, we develop a discrete state-space model that enables us to (1) characterize the decision-making strategies used by participants and (2) label trials according to the strategy used. Our model is a Hidden Markov Model (HMM) with multinomial Generalized Linear Models (GLMs) parameterizing the emission probabilities, which we refer to as the GLM-HMM. The model has K discrete internal states, each of which corresponds to a different mode of decision-making behavior. The animal or human’s choice on a particular trial is governed by state-specific GLMs, which describe the mapping from task covariates (sensory evidence as well as well as task-irrelevant variables like past choice and bias) to the decision. I will demonstrate that our model explains the choice behavior in these tasks considerably better than existing models, and I will discuss some of the retrieved decision-making states. As well as a useful tool for neuroscientists, we hope that enumerating and characterizing animal and human strategies may prove useful for computer scientists training artificial agents to perform decision-making tasks. Reading list: Textbook: Christopher M. Bishop, Pattern Recognition and Machine Learning (Springer, 2006). Papers: 1. Sean Escola et al., ‘Hidden Markov Models for the Stimulus-Response Relationships of Multistate Neural Systems’, Neural Computation 23, no. 5 (2011): 1071–1132. 2. Radford M. Neal and Geoffrey E. Hinton, ‘A View of the EM Algorithm That Justifies Incremental, Sparse, and Other Variants’, in Learning in Graphical Models (Springer, 1998), 355–368. 3. Ruslan Salakhutdinov, Sam T. Roweis, and Zoubin Ghahramani, ‘Optimization with EM and Expectation-Conjugate-Gradient’, in Proceedings of the 20th International Conference on Machine Learning (ICML-03), 2003, 672–679. 4. Naonori Ueda and Ryohei Nakano, ‘Deterministic Annealing EM Algorithm’, Neural Networks 11, no. 2 (1998): 271–282. 5. Zoubin Ghahramani, ‘An Introduction to Hidden Markov Models and Bayesian Networks’, in Hidden Markov Models: Applications in Computer Vision (World Scientific, 2001), 9–41. 6. Gianluigi Mongillo and Sophie Deneve, ‘Online Learning with Hidden Markov Models’, Neural Computation 20, no. 7 (2008): 1706–1716. 7. Matthew J. Beal, Zoubin Ghahramani, and Carl E. Rasmussen, ‘The Infinite Hidden Markov Model’, in Advances in Neural Information Processing Systems, 2002, 577–584. 8. Scott W. Linderman et al., ‘Hierarchical Recurrent State Space Models Reveal Discrete and Continuous Dynamics of Neural Activity in C. Elegans’, BioRxiv, 2019, 621540. 9. Alexander B. Wiltschko et al., ‘Mapping Sub-Second Structure in Mouse Behavior’, Neuron 88, no. 6 (16 December 2015): 1121–35, https://doi.org/10.1016/j.neuron.2015.11.031.https://doi.org/10.1016/j.neur…. 10. Nicholas A. Roy et al., ‘Efficient Inference for Time-Varying Behavior during Learning’, in Advances in Neural Information Processing Systems, 2018, 5695–5705.

1 0

Colloquium Speaker Dougal Maclaurin: Tuesday, Oct 15
by Emily C. Lawrence 09 Oct '19

09 Oct '19

CS Colloquium Speaker Dougal Maclaurin from Google Brain Tuesday, October 15 - 12:30pm Computer Science - Room 105 Host: Ryan Adams [ https://www.cs.princeton.edu/events/25877 | https://www.cs.princeton.edu/events/25877 ] JAX: Accelerated machine learning research via composable function transformations in Python JAX is a system for high-performance machine learning research. It offers the familiarity of Python+NumPy and the speed of hardware accelerators, and it enables the definition and the composition of function transformations useful for machine learning programs. In particular, these transformations include automatic differentiation, automatic batching, end-to-end-compilation (via XLA), and parallelizing over multiple accelerators. They are the key to JAX's power and to its relative simplicity. JAX had its initial open-source release in December 2018 (https://github.com/google/jax). It is currently being used by several groups of researchers for a wide range of advanced applications, from studying spectra of neural networks, to probabilistic programming and Monte Carlo methods, and scientific applications in physics and biology. Users appreciate JAX most of all for its ease of use and flexibility. Bio: Dougal Maclaurin is a research scientist at Google. He works on programming languages and systems for machine learning, particularly the Python library JAX. He started Autograd , a system for automatic differentiation in Python, which has inspired the design of several systems, including PyTorch , MinPy , Torch Autograd and Julia Autograd . He is a co-founder of Day Zero Diagnostics, a startup developing a sequencing-based diagnostic for drug-resistant infections. He received his Ph.D. from Harvard in 2016, working with Ryan Adams on the development of methods for machine learning. His work on scalable MCMC, "Firefly Monte Carlo", was recognized with the Best Paper award at UAI 2014.

1 0

Caroline Trippel will present her FPO, "Concurrency and Security Verification in Heterogeneous Parallel Systems" on Friday, 10/18/2019 at 9am in CS 105
by Nicki Mahler 08 Oct '19

08 Oct '19

Caroline Trippel will present her FPO, "Concurrency and Security Verification in Heterogeneous Parallel Systems" on Friday, 10/18/2019 at 9am in CS 105. The members of her committee are: Margaret Martonosi (Adviser); Readers: Aarti Gupta, Dan Lustig (NVIDIA), and Margaret Martonosi; Examiners: Andrew Appel, David Wentzlaff, and Margaret Martonosi. A copy of her thesis, is available upon request. Please email ngotsis(a)cs.princeton if you would like a copy of the thesis. Everyone is invited to attend her talk. The talk title and abstract follow below: To achieve performance scaling at manageable power and thermal levels, modern systems architects employ parallelism along with high degrees of hardware specialization and heterogeneity. Unfortunately, the power and performance improvements afforded by heterogeneous parallelism come at the cost of significantly increased design complexity, with different components being programmed differently and accessing shared resources differently. This design complexity in turn presents challenges for architects who need to devise mechanisms for orchestrating, enforcing, and verifying the correctness and security of executing applications. As it turns out, software-level correctness and security problems can result from problematic hardware event orderings and interleavings that take place when an application executes on a particular hardware implementation. Since hardware designs are complex, and since a single user-facing instruction can exhibit a variety of different hardware execution event sequences, analyzing and verifying systems for correct and secure orderings and interleavings of these events is challenging. To address this issue, this dissertation combines hardware systems architecture approaches with formal methods techniques to support the specification, analysis, and verification of implementation-aware event ordering scenarios. The specific goal here is enabling automatic synthesis of implementation-aware programs capable of violating correctness or security guarantees when such programs exist. First, this dissertation presents TriCheck, an approach and tool for conducting full-stack memory consistency model verification (from high-level programming languages down through hardware implementations). Using rigorous and ecient formal approaches, TriCheck identified flaws in the 2016 RISC-V memory model specification and two counterexamples to a previously proven-correct compiler mapping scheme from C11 onto Power and ARMv7. Second, after making the important observation that memory consistency model and security analyses are amenable to similar approaches, this thesis presents CheckMate, an approach and tool for conducting hardware security verification. CheckMate uses formal techniques to evaluate susceptibility of a hardware system design to formally-specified security exploit classes. When a design is susceptible, proof-ofconcept exploit codes are synthesized. CheckMate automatically synthesized programs representative of Meltdown and Spectre and new exploits, MeltdownPrime and SpectrePrime. Third, this dissertation presents approaches for handling memory model heterogeneity in hardware systems, focusing on correctness and highlighting applicability of the proposed techniques to security.

1 0