- talks - lists.cs.princeton.edu

14th Annual Innovation Forum
by Emily Lawrence 22 Feb '19

22 Feb '19

The 14th Annual Innovation Forum is fast approaching. The forum will take place on Wednesday, February 27th at 2:00pm in the Andlinger Center's Meader Hall at Princeton University. This event will showcase Princeton research developed by faculty, postdoctoral researchers and graduate students that offers the potential to be commercialized. Nine teams will be vying for prize money totally $30,000 to bring their innovations to market. Find out who is in the running here <https://kellercenter.princeton.edu/engage/innovation-forum/teams> and register here <https://princeton.us8.list-manage.com/track/click?u=7fbf3b0a07b0fd95cded6d6 03&id=41a8afb45c&e=44e890d167> to save your seat for this exciting chance to see the newest innovations emerging from Princeton University labs. Hosted by the Keller Center <https://kellercenter.princeton.edu/> and the Office of Technology Licensing <https://patents.princeton.edu/> Beth Ellen Jarvie Communication Coordinator Keller Center School of Engineering and Appied Science Princeton University ACE 25 Engineering Quadrangle Princeton, NJ 08544 609-258-1643

1 0

Andy Zeng will present his Pre FPO "Visual Affordances for Robotic Manipulation" on Tuesday, February 26, 2019 at 10am in CS 401
by Nicki Mahler 22 Feb '19

22 Feb '19

Andy Zeng will present his Pre FPO "Visual Affordances for Robotic Manipulation" on Tuesday, February 26, 2019 at 10am in CS 401. The members of his committee are as follows: Tom Funkhouser (adviser), Szymon Rusinkiewicz, Olga Russakovsky, Adam Finkelstein, Alberto Rodriguez (MIT). All are welcome to attend. Please see below for talk title and abstract. Title: Learning Visual Affordances for Robotic Manipulation Abstract: How do we use machine learning algorithms to enable robots to intelligently interact with the physical world? On the one hand, we can adapt the classic sense-plan-act paradigm by replacing individual modules (e.g. pose estimation) with deep networks -- but these algorithms remain specific to training scenarios and suffer from error propagation. On the other hand, we can learn end-to-end models that map from raw sensory data (RGB images) to low-level control (motor torques) -- but these models are highly data inefficient, and robot data is expensive. In this talk, I will propose a middleground: learning the visual affordances of motion primitives. The idea is to use stable control algorithms to design motion primitives, then use deep learning to map from visual observations to the perceived affordances (e.g. confidence scores, or action-values) of these primitives. By abstracting away low-level control from the learning, this formulation significantly improves data efficiency, while exhibiting versatility and generalization. First, I will show how this formulation can be used to learn manipulation policies for multiple grasping primitives that generalize to new objects, enabling a real-world picking system to win 1st place (stow task) at the Amazon Robotics Challenge. Next, I will discuss how we can adapt this framework to enable long-term planning for sequencing manipulation primitives (e.g. pushing and grasping) with self-supervised deep reinforcement learning. This results in a system that learns to execute complex manipulation strategies that can generalize to new objects, and requires orders of magnitude less data than prior work. Lastly, I will talk about how we can augment this formulation by inferring additional control parameters -- in particular, by learning to predict residuals on top of parameter estimates from physics-based control. This enables the creation of TossingBot, a robot arm that is capable of grasping arbitrary objects and throwing them to target locations outside its maximum reach range. TossingBot throws more accurately than the average Google engineer, learns within a few hours via self-supervision, generalizes to new objects and locations, and achieves picking speeds twice as fast as state-of-the-art systems at 500+ mean picks per hour. In all settings, our experiments demonstrate that learning affordances for motion primitives can enable systems to quickly learn complex manipulation skills that scale to novel scenarios.

1 0

Colloquium Speaker Elissa Redmiles TODAY at 12:30pm
by Emily C. Lawrence 21 Feb '19

21 Feb '19

Colloquium Speaker Elissa Redmiles from University of Maryland Thursday, February 21 - 12:30pm Computer Science - Room 105 Host: Arvind Narayanan [ http://www.cs.princeton.edu/events/25817 | http://www.cs.princeton.edu/events/25817 ] Security for All: Modeling Structural Inequities to Design More Secure Systems Users often fall for phishing emails, reuse simple passwords, and fail to effectively utilize "provably" secure systems. These behaviors expose users to significant harm and frustrate industry practitioners and security researchers alike. As consequences of security breaches become ever more grave, it is important to study why humans behave seemingly irrationally. In this talk, I will illustrate how modeling the effects of structural inequities -- variance in skill, socioeconomic status, as well as culture and gender identity -- can both explain apparent irrationality in users’ security behavior and offer tangible improvements in industry systems. Modeling and mitigating security inequities requires a combination of techniques from economic, data scientific, and social science methodologies to develop new tools for systematically understanding and mitigating insecure behavior. Through novel experimental methodology, I empirically show strong evidence of bounded rationality in security behavior: Users make mathematically modelable tradeoffs between the protection offered by security behaviors and the costs of practicing those behaviors, which even in a highly usable system may outweigh the benefits, especially for less resourced users. These findings emphasize the need for industry systems that balance structural inequities and accommodate behavioral variance between users rather than one-size-fits-all security solutions. More broadly, my techniques for modeling and accounting for inequities have offered key insights in growing technical areas beyond security, including algorithmic fairness. Bio: Elissa Redmiles is a Ph.D. Candidate in Computer Science at the University of Maryland and has been a visiting researcher with the Max Planck Institute for Software Systems and the University of Zurich. Elissa’s research interests are broadly in the areas of security and privacy. She uses computational, economic, and social science methods to conduct research on behavioral security. Elissa seeks to understand users’ security and privacy decision-making processes and specifically investigate inequalities that arise in these processes and to mitigate those inequalities through the design of systems that facilitate safety equitably across users. Elissa is the recipient of a NSF Graduate Research Fellowship, a National Science Defense and Engineering Graduate Fellowship, and a Facebook Fellowship. Her work has appeared in popular press publications such as Scientific American, Business Insider, Newsweek, and CNET and has been recognized with the John Karat Usable Privacy and Security Student Research Award, a Distinguished Paper Award at USENIX Security 2018, and a University of Maryland Outstanding Graduate Student Award.

1 0

EE and CS Speaker-Jaime Fernández Fisac, 2/26 at 4:30pm
by Emily C. Lawrence 21 Feb '19

21 Feb '19

EE and CS Speaker Jaime Fernández Fisac from the University of California, Berkeley Tuesday, February 26 - 4:30pm Computer Science 105 Hosts: Naveen Verma , EE and Szymon Rusinkiewicz, CS [ http://ee.princeton.edu/events/resilient-safety-assurance-robotic-systems-s… | http://ee.princeton.edu/events/resilient-safety-assurance-robotic-systems-s… ] Resilient Safety Assurance for Robotic Systems: Staying Safe Even When Models Are Wrong In order for autonomous systems like robots, drones, and self-driving cars to be reliably introduced into our society, they must be able to actively account for safety during their operation. While safety analysis has traditionally been conducted offline for controlled environments like cages on factory floors, the much higher complexity of open, human-populated spaces like our homes, cities, and roads means that any precomputed guarantees may become invalid when modeling assumptions made at design time are violated once the system is deployed. My research aims to enable autonomous systems to proactively ensure safety during their operation by explicitly reasoning about the gap between their models and the real world. In this talk I will present recent contributions to safety assurance for autonomous systems. I will first discuss new advances in efficient safety computation, and demonstrate their use in large-scale unmanned air traffic. Next, I will present a general safety framework enabling the use of learning control schemes (e.g. reinforcement learning) for safety-critical robotic systems in uncertain environments. I will then turn my attention to the important problem of safe human-robot interaction, and introduce a real-time Bayesian method to monitor the reliability of predictive human models. The talk will end with a discussion of challenges and opportunities ahead, including the introduction of game-theoretic planning in autonomous driving and the bridging of safety analysis and deep reinforcement learning. Bio: Jaime Fernández Fisac is a Ph.D. candidate in Electrical Engineering and Computer Sciences at UC Berkeley. His research interests lie in control theory and artificial intelligence, with a focus on safety assurance for autonomous systems. He works to enable robotic systems to safely interact with uncertain environments and human beings despite using inaccurate models. Jaime received a B.S./M.S. degree in Electrical Engineering from the Universidad Politécnica de Madrid, Spain, in 2012, and a M.Sc. in Aeronautics from Cranfield University, U.K., in 2013. He is a recipient of the La Caixa Foundation fellowship.

1 0

Reminder: Colloquium Speaker Elissa Redmiles tomorrow at 12:30pm
by Emily C. Lawrence 20 Feb '19

20 Feb '19

Colloquium Speaker Elissa Redmiles from University of Maryland Thursday, February 21 - 12:30pm Computer Science - Room 105 Host: Arvind Narayanan [ http://www.cs.princeton.edu/events/25817 | http://www.cs.princeton.edu/events/25817 ] Security for All: Modeling Structural Inequities to Design More Secure Systems Users often fall for phishing emails, reuse simple passwords, and fail to effectively utilize "provably" secure systems. These behaviors expose users to significant harm and frustrate industry practitioners and security researchers alike. As consequences of security breaches become ever more grave, it is important to study why humans behave seemingly irrationally. In this talk, I will illustrate how modeling the effects of structural inequities -- variance in skill, socioeconomic status, as well as culture and gender identity -- can both explain apparent irrationality in users’ security behavior and offer tangible improvements in industry systems. Modeling and mitigating security inequities requires a combination of techniques from economic, data scientific, and social science methodologies to develop new tools for systematically understanding and mitigating insecure behavior. Through novel experimental methodology, I empirically show strong evidence of bounded rationality in security behavior: Users make mathematically modelable tradeoffs between the protection offered by security behaviors and the costs of practicing those behaviors, which even in a highly usable system may outweigh the benefits, especially for less resourced users. These findings emphasize the need for industry systems that balance structural inequities and accommodate behavioral variance between users rather than one-size-fits-all security solutions. More broadly, my techniques for modeling and accounting for inequities have offered key insights in growing technical areas beyond security, including algorithmic fairness. Bio: Elissa Redmiles is a Ph.D. Candidate in Computer Science at the University of Maryland and has been a visiting researcher with the Max Planck Institute for Software Systems and the University of Zurich. Elissa’s research interests are broadly in the areas of security and privacy. She uses computational, economic, and social science methods to conduct research on behavioral security. Elissa seeks to understand users’ security and privacy decision-making processes and specifically investigate inequalities that arise in these processes and to mitigate those inequalities through the design of systems that facilitate safety equitably across users. Elissa is the recipient of a NSF Graduate Research Fellowship, a National Science Defense and Engineering Graduate Fellowship, and a Facebook Fellowship. Her work has appeared in popular press publications such as Scientific American, Business Insider, Newsweek, and CNET and has been recognized with the John Karat Usable Privacy and Security Student Research Award, a Distinguished Paper Award at USENIX Security 2018, and a University of Maryland Outstanding Graduate Student Award.

1 0

Sachin Ravi will present his Pre FPO "Meta-Learning for Data and Processing Efficiency" on Tuesday, February 26, 2019 at 9am in CS 301.
by Nicki Mahler 19 Feb '19

19 Feb '19

Sachin Ravi will present his Pre FPO "Meta-Learning for Data and Processing Efficiency" on Tuesday, February 26, 2019 at 9am in CS 301. The members of his committee are as follows: Kai Li (adviser), Jonathan Cohen (PNI), Hugo Larochelle (Google Brain), Jia Deng, and Karthik Narasimhan. All are welcome to attend. Please see below for talk title and abstract. Title : Meta-Learning for Data and Processing Efficiency Abstract : Deep learning models have shown great success in a variety of areas, including computer vision, natural language processing, and reinforcement learning. However, deep learning methods still lack the efficiency and flexibility of human learning. Current deep learning methods involve training on a large amount of data to produce a model that can then specialize to the specific task encoded by the training data. Humans, on the other hand, are constantly learning and mastering new concepts throughout our lives with comparatively little experience. In order to bridge this gap, previous work has suggested the use of meta-learning. Rather than learning how to do a specific task, meta-learning involves learning how-to-learn and utilizing this knowledge to learn new tasks more effectively. This talk focuses on work using meta-learning to improve the data and processing efficiency of deep learning models when learning new tasks. In the first part, we discuss a meta-learning model for the few-shot learning problem, where the aim is to be able to learn a new classification task having unseen classes with few labeled examples. We use a LSTM-based meta-learner model to learn both the initialization and the optimization algorithm used to train another neural network in the few-shot regime and show that our method compares favorably to nearest-neighbor approaches. The second part of the thesis deals with improving the predictive uncertainty of models in the few-shot learning setting. Proper predictive uncertainty is key to deploying machine learning models in the wild (especially in settings with little data), as proper human intervention can be applied when a model’s prediction is known to be uncertain. Using a Bayesian perspective, we propose a meta-learning method which efficiently amortizes hierarchical variational inference across tasks, learning a prior distribution over neural network weights so that a few steps of gradient descent will produce a good task-specific approximate posterior. Finally, we focus on applying meta-learning in the context of making choices that impact processing efficacy. When training a network on multiple tasks, we have a choice between interactive parallelism (training on different tasks one after another) and independent parallelism (using the network to process multiple tasks concurrently). For the simulation environment we consider, we show that there is a trade-off between these two types of processing choices in deep neural networks, where one can learn faster by using interactive parallelism but at the cost of having to do tasks serially. We then discuss a meta-learning algorithm for an agent to learn how to train itself with regard to this trade-off in an environment with unknown serialization cost.

1 0

Yatin Manerkar will present his Pre-FPO "Hierarchical Memory Consistency Verification of Parallel System Implementations" on Tuesday, February 26th at 10am in CS 302
by Nicki Mahler 19 Feb '19

19 Feb '19

Yatin Manerkar will present his Pre-FPO on Tuesday, February 26, 2019 at 10am in CS 302. The members of his committee are as follows: Margaret Martonosi (Adviser), Aarti Gupta (reader), Daniel Lustig (NVIDIA, reader), Sharad Malik (EE, non-reader), and Zachary Kincaid (non-reader) All are welcome to attend. The talk title and abstract follow below. Title: Hierarchical Memory Consistency Verification of Parallel System Implementations Abstract: With the end of Moore's Law and Dennard scaling, system developers have turned to parallel architectures and languages to continue to improve performance. Memory consistency models (MCMs) specify the ordering requirements for memory operations used for synchronization in these parallel systems, so MCM verification is critical to overall system correctness. My work enables efficient formal MCM modelling and verification across the stack, from high-level languages down to Verilog RTL. In this talk, I will discuss two components of my thesis: PipeProof and RTLCheck. PipeProof offers the first-ever automated verification of a hardware design specification against its instruction-level MCM across all possible programs (an infinite space). PipeProof uses a combination of abstraction refinement and happens-before graphs that are checked using an SMT (Satisfiability Modulo Theories) solver, and enables hardware architects to automatically ensure that their design respects its MCM before even a line of RTL (Register Transfer Language) is implemented. Once the processor's low-level Verilog RTL exists, the RTLCheck approach I developed can automatically check this Verilog against the hardware specification verified by PipeProof for a set of test programs. A key contribution of RTLCheck is its set of novel algorithms for connecting the temporal execution of RTL to the axiomatic happens-before graphs used by PipeProof. The composition of PipeProof and RTLCheck allows designers to automatically check their low-level RTL against an instruction-level MCM specification.

1 0

TODAY at 4:30pm in B205 E-Quad - Petter Nillson, California Institute of Technology
by Emily Lawrence 18 Feb '19

18 Feb '19

EE SEMINAR SERIES Speaker: Petter Nilsson, California Institute of Technology Title: Specification-Driven Design for Modular and Safe Robotics Day: Monday, February 18, 2019 Time: 4:30 pm Room: B205 Engineering Quadrangle Host: Prof. Gerard Wysocki Robotic systems of tomorrow will be increasingly interconnected and operate amongst us, which implies a two-fold engineering challenge of great complexity and no tolerance for mistakes. This talk will explore specification-driven design methods that enforce or utilize formally written specifications for principled design, modularity, and decision-making. The first part will be centered on safety-critical control via invariance: I will show how invariance specifications in the form of assume-guarantee contracts can be leveraged to decompose problems and thus enable modular design, and how certificates for invariance can be used to formally relate low-level dynamics to a high-level abstract roadmap for planning. The second part of the talk will cover specification-guided methods for multi-robot systems, and how problem structure can be leveraged to overcome scalability challenges. The talk will be concluded with a few words about current research topics and directions for the future. Bio: Petter Nilsson received his B.S. in Engineering Physics in 2011, and his M.S. in Optimization and Systems Theory in 2013, both from KTH Royal Institute of Technology in Stockholm, Sweden, and his Ph.D. in Electrical Engineering in 2017 from the University of Michigan. In addition to his technical degrees, he holds a B.S. in Business and Economics from the Stockholm School of Economics. He is currently a postdoctoral scholar at the California Institute of Technology where he conducts research on specification-driven control and autonomy for safety-critical cyber-physical systems, with applications in autonomous driving, space exploration, and multi-agent coordination.

1 0

Logan Stafman will be presenting his Pre FPO on Friday, February 22, 2019 at 3pm in CS 302.
by Nicki Mahler 15 Feb '19

15 Feb '19

Logan Stafman will be presenting his Pre FPO on Friday, February 22, 2019 at 3pm in CS 302. The members of his committee are as follows: Michael Freedman (adviser), Wyatt Lloyd, Kyle Jamieson, Amit Levy, and Jennifer Rexford. All are welcome to attend. Please see below for talk title and abstract. Utility Scheduling in Multi-Tenant Clusters The rapid increase in data size along with the complex patterns of data usage amongst data scientists presents new challenges for large-scale data analytics systems. Modern distributed computing frameworks must support complex applications that range from answering database queries to training machine learning models. As data centers have grown, managing their resources has become an increasingly important task. New applications have become popular that make traditional scheduling systems inadequate. In this thesis, we present distributed scheduling systems aimed at increasing cluster resource utilization by taking advantage of specific characteristics of data processing applications. We present the following systems: (i) SLAQ: a cluster scheduling system for machine learning (ML) training jobs that aims to maximize the qualities of all models trained. In exploratory model training, models can be improved more quickly by redirecting resources to jobs with the highest potential for improvement. SLAQ reduces latency and maximizes the quality of models being trained by a distributed ML cluster. (ii) ReLAQS: a cluster scheduling system for incremental approximate query processing (AQP) systems that aims to minimize the error of all approximate results. In AQP, queries compute approximate results by sampling data. In AQP, error can be reduced more quickly by allocating resources to queries with higher error. ReLAQS reduces the latency required to reach a query result with a given level of error in a shared AQP environment. These works demonstrate a novel set of methods that can be used in fine-grained scheduling to build responsive, efficient distributed systems. We have evaluated these systems on standard benchmark workloads and datasets, as well as popular ML algorithms, and show both reduced latency and increased accuracy of intermediary results.

1 0

Colloquium Speaker-Elissa Redmiles, Thursday Feb 21
by Emily C. Lawrence 15 Feb '19

15 Feb '19

Colloquium Speaker Elissa Redmiles from University of Maryland Thursday, February 21 - 12:30pm Computer Science - Room 105 Host: Arvind Narayanan [ http://www.cs.princeton.edu/events/25817 | http://www.cs.princeton.edu/events/25817 ] Security for All: Modeling Structural Inequities to Design More Secure Systems Users often fall for phishing emails, reuse simple passwords, and fail to effectively utilize "provably" secure systems. These behaviors expose users to significant harm and frustrate industry practitioners and security researchers alike. As consequences of security breaches become ever more grave, it is important to study why humans behave seemingly irrationally. In this talk, I will illustrate how modeling the effects of structural inequities -- variance in skill, socioeconomic status, as well as culture and gender identity -- can both explain apparent irrationality in users’ security behavior and offer tangible improvements in industry systems. Modeling and mitigating security inequities requires a combination of techniques from economic, data scientific, and social science methodologies to develop new tools for systematically understanding and mitigating insecure behavior. Through novel experimental methodology, I empirically show strong evidence of bounded rationality in security behavior: Users make mathematically modelable tradeoffs between the protection offered by security behaviors and the costs of practicing those behaviors, which even in a highly usable system may outweigh the benefits, especially for less resourced users. These findings emphasize the need for industry systems that balance structural inequities and accommodate behavioral variance between users rather than one-size-fits-all security solutions. More broadly, my techniques for modeling and accounting for inequities have offered key insights in growing technical areas beyond security, including algorithmic fairness. Bio: Elissa Redmiles is a Ph.D. Candidate in Computer Science at the University of Maryland and has been a visiting researcher with the Max Planck Institute for Software Systems and the University of Zurich. Elissa’s research interests are broadly in the areas of security and privacy. She uses computational, economic, and social science methods to conduct research on behavioral security. Elissa seeks to understand users’ security and privacy decision-making processes and specifically investigate inequalities that arise in these processes and to mitigate those inequalities through the design of systems that facilitate safety equitably across users. Elissa is the recipient of a NSF Graduate Research Fellowship, a National Science Defense and Engineering Graduate Fellowship, and a Facebook Fellowship. Her work has appeared in popular press publications such as Scientific American, Business Insider, Newsweek, and CNET and has been recognized with the John Karat Usable Privacy and Security Student Research Award, a Distinguished Paper Award at USENIX Security 2018, and a University of Maryland Outstanding Graduate Student Award.

1 0