CIS Seminars & Events

Fall 2017 Colloquium Series

Unless otherwise noted, our lectures are held weekly on Tuesday or Thursday from 3:00 p.m. to 4:15 p.m. in Wu and Chen Auditorium, Levine Hall.

September 7th

Hovav Shacham
Department of Computer Science and Engineering
University of California, San Diego
"Trusted Browsers for Uncertain Times"

Read Abstract and Bio

Abstract

The Web is the most important platform for human communication, including communication by human-rights activists, reporters, dissidents, and others at risk of reprisal for their speech. Modern browsers are designed to keep their users’ computers from being compromised in “browse-by” attacks, they have no comparable architectural protections for users’ privacy. In this talk, I will show how the unintended interactions of browser features with underlying system components can be used — and are used — by malicious sites to track and identify Web users and to learn information shared by those users with other sites, in violation of the browser’s intended compartmentalization guarantees. Two decades of experience show that vendors’ ad-hoc efforts to close each hole as it is identified are unlikely to produce a browser that keeps its users’ secrets. Drawing on the trusted systems literature of the 1980s, I present a principled browser design that provably reduces the capacity of all timing channels that expose secret information.

Bio
Hovav Shacham is a Professor of Computer Science and Engineering at the University of California, San Diego. His research interests are in applied cryptography, systems security, privacy-enhancing technologies, and technology policy. Shacham was a student at Stanford and a postdoctoral fellow at the Weizmann Institute. He took part in California’s 2007 “Top-to-Bottom” voting systems review and served on the advisory committee for California’s 2011–13 post-election risk-limiting audit pilot program. His work has been cited by the Federal Trade Commission, the National Highway Traffic Safety Administration, and the RAND Corporation.

September 12th

Michael Franklin
Department of Computer Science
University of Chicago
"Big Data Software: What’s Next?"

Read Abstract and Bio

Abstract

The Big Data revolution has been enabled by a wealth of innovation in software platforms for data storage, analytics, and machine learning. The design of Big Data platforms such as Hadoop and Spark focused on scalability, fault-tolerance and performance. As these and other systems increasingly become part of the mainstream, the next set of challenges are becoming clearer. Requirements for performance are changing as workloads and hardware evolve. But more fundamentally, other issues are moving to the forefront. These include ease of use for a wide range of users, security, concerns about privacy and potential bias, and the perennial problems of data quality and integration from heterogeneous sources. In this talk, I will give an overview of how we got here, with an emphasis on the development of the Apache Spark system. I will then focus on these emerging issues with an eye towards where the academic research community can most effectively engage.

Bio
Michael Franklin is the Liew Family Chair of Computer Science at the University of Chicago where he also serves as senior advisor to the provost on computation and data science. Previously he was at UC Berkeley where he was the Thomas M. Siebel Professor of Computer Science and Chair of the Computer Science Division. He co-founded Berkeley’s AMPLab, a leading academic big data analytics research center, and served as an executive committee member for the Berkeley Institute for Data Science, a campus-wide initiative to advance data science environments. Michael is a Fellow of ACM, a two-time recipient of the ACM SIGMOD “Test of Time” award, and the Outstanding Advisor award from the Berkeley CS Graduate
Student Association.

 

September 21st

Special CIS Lecture
Gaurav Chakravorty
Co-Founder, CIO
qplum
"The next ten years: Deep Learning in Trading"

Read Abstract and Bio

Abstract

Gaurav will talk about some job trends in financial services to start and cover the transformational impact of A.I. and Deep Learning in making trading a scientific process. Deep learning has been very successful in social sciences and specially areas where there is a lot of data. Trading and investment management fits this paradigm perfectly. It is a social science and not a pure science, and we are generating petabytes of data everyday making it tough to learn from. With the advent of Deep Learning and Big Data technologies for efficient computation, we are finally able to use the same methods in investment management as we would in face recognition or making chatbots. This focus on learning a hierarchical set of concepts is truly making investing a scientific process, a utility.

Bio
Gaurav Chakravorty is a UPenn alumnus (MSE CIS '05) and co-founder and CIO at qplum. Qplum is an asset management firm that offers A.I. based trading strategies. Gaurav has been one of the early pioneers in machine learning based high-frequency trading. He built the most profitable algo trading group at Tower Research from 2005-2010 and was the youngest partner in the firm. Gaurav's strategies have made more than $1.4bln to-date. He believes in the potential of using Deep Learning to reduce fees and make investing a science that is universally accessible.

 

September 28th

John Hughes
Computing Science Department
Chalmers University of Technology
"The Incredible Shrinking Test Case"

Read Abstract and Bio

Abstract

QuickCheck is a random testing tool that not only tests code against its specification, but also "shrinks" any failing tests to minimal counterexamples. Since 2006, Quviq has been helping customers apply this idea to industrial software. In this talk, I'll focus on the value of shrinking, showing how it can be used to simplify fault diagnosis, to avoid repeatedly testing for the same bugs, to refine the idea of "testíng a requirement" and to illustrate specifications with examples.

Bio
John Hughes has worked in the area of functional programming since around 1980, authoring one of the most widely read introductions to the area, "Why Functional Programming Matters", and helping to design Haskell. In 2000 he and Koen Claessen published "QuickCheck: A Lightweight Random Testing Tool for Haskell" at ICFP, which in 2010 received the "Most Influential Paper" award for that year. In 2006 he and Thomas Arts founded Quviq, a start-up based on the QuickCheck idea, and since then he has divided his time between Quviq and Chalmers University in Gothenburg, Sweden.

 

October 3rd

Salvatore Stolfo
Department of Computer Science
Columbia University
"A Brief History of Symbiote Defense"

Read Abstract and Bio

Abstract

Market watchers estimate the IoT Security marketplace is now valued at over $6 Billion and expected to reach $22 Billion by 2020. Just 5 years ago, embedded device security was barely on the map. Our early work in the IDS Lab at Columbia demonstrated the seriousness of the embedded device insecurity problem, and the relatively easy exploitation of devices such as printers, IP phones and routers. Recent advances in offensive technologies targeting a wide range of IoT devices have shown that the exploitation of these lucrative but poorly designed devices is not terribly difficult, including medical products, SCADA devices, automobiles and refrigerators. The goal of our early work was to defend embedded devices with an entirely new defensive capability we call the Software Symbiote, a host-based defensive technology that automatically injects intrusion detection functionality within the firmware of any device. In this talk we will provide a brief history of our work on the Symbiote technology, and the transition from academic research to practical and wide-spread use in common commodity products.

Bio
Salvatore Stolfo is a Professor of Computer Science at Columbia University. He is regarded as creating the area of machine learning applied to computer security in the mid-1990’s and has created several anomaly detection algorithms and systems addressing some of the hardest problems in securing computer systems. Of particular note is his recent interest in the practical application of deception security in scale. Stolfo is also co-inventor of the Symbiote technology that automatically injects intrusion detection functionality into arbitrary embedded devices. Stolfo has had numerous best papers and awards, most recently the RAID Most Influential Paper and Usenix Security Distinguished Paper awards. He has published well over 230 papers and has been granted over 60 patents and has been an advisor and consultant to government agencies, including DARPA, the National Academies and others, for well over 2 decades. Two security companies were recently spun out of his IDS lab, Allure Security Technology and Red Balloon Security.

 

October 12th

Arnab Nandi
Department of Computer Science and Engineering
Ohio State University
"Chasing Interactivity: Querying Beyond Keyboards"

Read Abstract and Bio

Abstract

New computing interfaces that use "natural" modes of interaction — such as multitouch and gestures — are rapidly becoming more popular than traditional keyboard-based interaction. These devices are being used to consume and directly interact with data in a wide range of contexts, from business intelligence to data-driven sciences. Applications for such devices are highly interactive, and pose a fundamentally different set of expectations on the underlying data infrastructure. In this talk, we rethink various aspects of the data infrastructure stack, from the query specification process to distributed query execution, to address interactive workloads. We explore the impact of including interactivity as first-class concept, and show that our methods result in experiences that are not only fluid, but also more intuitive for the end-user.

Bio
Arnab Nandi is an Associate Professor in the Computer Science and Engineering department at The Ohio State University. Arnab’s research is in the area of database systems, focusing challenges in large-scale data analytics and human-in-the-loop data exploration. Arnab is also a founder of The STEAM Factory, a collaborative interdisciplinary research and public outreach initiative, and faculty director of the OHI/O Informal Learning Program. Arnab is a recipient of the US National Science Foundation’s CAREER Award, a Google Faculty Research Award, and the Ohio State College of Engineering Lumley Research Award. He is also the 2016 recipient of the IEEE TCDE Early Career Award for his contributions towards user-focused data interaction.

 

October 16th

CIS Lecturer Candidate
Adam Blank
Computer Science and Engineering
University of Washington
"Structure and Application of AVL Trees"

Read Abstract and Bio

Abstract
Binary Search Trees (BSTs) are an efficient data structure for storing information in a naturally sorted order. In order to take advantage of the tree structure, though, the tree has to be able to balance itself in order to minimize its height. This sample lecture would be part of a data structures class in Java. We will assume that the course has already covered runtime analysis, binary trees, and binary search trees. This lecture will discuss AVL Trees, which are the first type of BSTs that students are likely to see.

Bio
Adam Blank is currently a Lecturer in the Paul G. Allen School of Computer Science at University of Washington. He received his M.S. in Computer Science from Carnegie Mellon University in Spring 2014. He is interested in using technology, collaboration, experimentation, and active learning to help CS college students get as much as they can out of their education.

 

October 26th

Avrim Blum
Professor and Chief Academic Officer
The Toyota Technological Institute at Chicago
"Learning about Agents and Mechanisms from Opaque Transactions"

Read Abstract and Bio

Abstract

In this talk I will discuss the problem of trying to learn the requirements and preferences of economic agents by observing the outcomes of an allocation mechanism whose rules you also don’t initially know. As an example, consider observing web pages where the agents are advertisers and the winners are those whose ads show up on the given page. We know these ads are placed based on bids and other constraints given to some auction mechanism, but we do not get to see these bids and constraints. What we would like to do is from repeated observations of this type to learn what the requirements and preferences of the agents are. Or consider observing the input-output behavior of a cloud computing service, where the input consists of a set of agents requesting service, and the output tells us which actually received service and which did not. In this case, we assume the agents who did not receive service were not served due to overlap of their resource needs with higher-priority requests. From such input-output behavior, we would like to learn the underlying structure. Our goal will be from observing a series of such interactions to try to learn both the needs and preferences of the agents and perhaps also the rules of the allocation mechanism.

This talk is based on work joint with Yishay Mansour and Jamie Morgenstern, as well as work joint with Michael Liang.

Bio
Avrim Blum received his BS, MS, and PhD from MIT in 1987, 1989, and 1991 respectively. He then served on the faculty in the Computer Science Department at Carnegie Mellon University from 1992 to 2017. In 2017 he joined the Toyota Technological Institute at Chicago as Chief Academic Officer.

Prof. Blum’s main research interests are in Theoretical Computer Science and Machine Learning, including Machine Learning Theory, Approximation Algorithms, Algorithmic Game Theory, and Database Privacy, as well as connections among them. Some current specific interests include multi-agent learning, multi-task learning, semi-supervised learning, and the design of incentive systems. He is also known for his past work in AI Planning. Prof. Blum has served as Program Chair for the IEEE Symposium on Foundations of Computer Science (FOCS) and the Conference on Learning Theory (COLT). He has served as Chair of the ACM SIGACT Committee for the Advancement of Theoretical Computer Science and on the SIGACT Executive Committee. Prof. Blum is recipient of the AI Journal Classic Paper Award, the ICML/COLT 10-Year Best Paper Award, the Sloan Fellowship, the NSF National Young Investigator Award, and the Herbert Simon Teaching Award, and he is a Fellow of the ACM.

 

November 2nd

Kunle Olukotun
Departments of Electrical Engineering and Computer Science
Stanford University
"Scaling Machine Learning Performance with Moore's Law"

Read Abstract and Bio

Abstract

Significant advances in Machine Learning (ML) have been enabled by powerful, efficient computing hardware. Today, this computing hardware is composed of CPUs and ML accelerators, which have traditionally been GPUs and in some instances FPGAs. Both of these accelerators are difficult to program. In this talk, I will describe a new design paradigm for ML accelerators that can make ML computing hardware both more efficient and simpler to program. The key to this new paradigm is to enable application developers to reconfigure hardware to match the requirements of their ML applications. The full-stack design paradigm consists of new ML algorithms, a new ML-specific programming language, new compilation technology and a new reconfigurable hardware architecture.

Bio
Kunle Olukotun is the Cadence Design Systems Professor of Electrical Engineering and Computer Science at Stanford University. Olukotun is well known as a pioneer in multicore processor design and the leader of the Stanford Hydra chip multipocessor (CMP) research project. Olukotun founded Afara Websystems to develop high-throughput, low-power multicore processors for server systems. The Afara multicore processor, called Niagara, was acquired by Sun Microsystems. Niagara derived processors now power all Oracle SPARC-based servers. Olukotun currently directs the Stanford Pervasive Parallelism Lab (PPL), which seeks to proliferate the use of heterogeneous parallelism in all application areas using Domain Specific Languages (DSLs). Olukotun is a member of the Data Analytics for What’s Next (DAWN) Lab which is developing infrastructure for usable machine learning. Olukotun is an ACM Fellow and IEEE Fellow for contributions to multiprocessors on a chip and multi-threaded processor design. Olukotun received his Ph.D. in Computer Engineering from The University of Michigan.

 

November 7th

Ehsan Hoque
Computer Science Department
University of Rochester
"When can a computer improve your social skills?"

Read Abstract and Bio

Abstract
Many people fear automation. They may see it as a potential job killer. They may also be concerned about what can be automated. Could we train a computer to teach us human skills? Should we? Artificial intelligence, when designed properly, can help people improve important social and cognitive skills. My research group has shown how automated systems can develop skills that improve performance in job interviews, public speaking, negotiations, working as part of a team, producing vowels during music
training, end-of-life communication between oncologists and cancer patients, and even routine social
interactions for people with Asperger’s syndrome. In this talk, I will offer insights gained from our exploration of several questions: How are humans able to improve important social and cognitive skills with a computer? What aspect of the feedback helps the most? How to design experiments to ensure that the skills generalize?

Bio
M. Ehsan Hoque is an assistant professor of computer science at the University of Rochester, where he leads the Rochester Human-Computer Interaction, or ROC HCI, Group. His group’s research focuses on understanding and modeling unwritten rules of human communication with applications in business communication, health, and assessment technologies. Ehsan received his Ph.D. from the MIT Media Lab in 2013 where his dissertation work was highlighted by the MIT Museum as one of the most unconventional inventions at MIT. Ehsan and his group’s work has received a Best Paper Award at Ubiquitous Computing (UbiComp 2013), Best Paper Honorable Mentions in Automated Face and Gesture Recognition (FG 2011) and Intelligent Virtual Agents (IVA 2006). Ehsan has received MIT TR35 Award(2016), World Technology Award (2016), and Google Faculty Award (2014, 2016). In 2017, Science News recognized him as one of 10 early to mid-career scientists to watch (the SN 10). Follow the group’s work on Twitter at @rochci.

 

November 9th

Konrad Kording
Department of Neuroscience, Department of Bioengineering
UPenn PIK Professor
"Rethinking the role of machine learning in (neuro)science"

Read Abstract and Bio

Abstract

The goal of much of computational biology is to numerically describe data from a system, but also to find ways of fixing it and to understand a system’s objectives, algorithms, and mechanisms. Here we will argue that, regardless the objective, machine learning should be a central contribution to progress in every flavor of biomedical science. Machine learning can typically better describe the data. In doing so it can also provide a benchmark for any other way of describing the data. Using examples from neuroscience we discuss how better performance matters for decoding models and how having a benchmark affects encoding models. Similar issues matter in medicine. As biomedical science evolves, machine learning is morphing into a critical tool across the full spectrum of scientific questions.

Bio
Konrad Kording is a Professor at the University of Pennsylvania in the Departments of Bioengineering and Neuroscience. He received his PhD in Physics from ETH Zurich and did postdoctoral training at ETH Zurich, UCL London, and MIT focusing on statistical approaches to the brain and cognition. He is using data science towards understanding the brain, medicine, and scientists themselves.

November 16th

Grace Hopper Lecture Series
Deborah Estrin
Department of Computer Science
Cornell Tech
"In pursuit of digital biomarkers"

Read Abstract and Bio

Abstract
Social networks, search engines, mobile apps, IoT vendors, online entertainment, and e-commerce sites have lead the way in using an individual’s digital traces to tailor service offerings, improve system performance, and target advertisements. A growing community of researchers are looking to these same data sources to create digital biomarkers for use in personalized health and wellness applications. This talk will discuss motivation and progress.

Bio
Deborah Estrin (PhD, MIT; BS, UCB) is a Professor of Computer Science at Cornell Tech in New York City where she founded the Jacobs Institute’s Health Tech Hub. She is also a co-founder of the non-profit startup, Open mHealth. Her current focus is on mobile health and small data, leveraging the pervasiveness of mobile devices and digital interactions for health and life management (TEDMED http://smalldata.io). Previously, Estrin was the founding director of the NSF-funded Science and Technology Center for Embedded Networked Sensing (CENS) at UCLA (2002-12). Estrin is an elected member of the American Academy of Arts and Sciences (2007) and National Academy of Engineering (2009). She was awarded honorary doctorates from EPFL and Uppsala.

 

November 30th

Aditya Parameswaran
Computer Science Department
University of Illinois at Urbana-Champaign
"Enabling Data Science for the 99%"

Read Abstract and Bio

Abstract

There is a severe lack of interactive tools to help people manage, analyze, and make sense of large datasets. This talk will briefly cover three tools under development in our research group (with collaborators at Illinois, MIT, Maryland, and Chicago) that empower individuals and teams to perform interactive data analysis more effectively. The three tools span the spectrum of analyses types --- from browsing with DataSpread, a spreadsheet-database hybrid, to exploration with ZenVisage, a effortless visualization recommendation tool, and finally to analysis and collaboration with Orpheus, a database system that supports versioning as a first-class citizen.

Bio
Aditya Parameswaran is an Assistant Professor in Computer Science at the University of Illinois (UIUC). He spent a year as a PostDoc at MIT CSAIL following his PhD at Stanford University, before starting at Illinois in August 2014. He develops systems and algorithms for "human-in-the-loop" data analytics, synthesizing techniques from database systems, data mining, and human computation. Aditya received the NSF CAREER Award, the TCDE Early Career Award, the C. W. Gear Junior Faculty Award from Illinois, multiple "best" Doctoral Dissertation Awards (from SIGMOD, SIGKDD, and Stanford), an "Excellent" Lecturer award from Illinois, a Google Faculty award, the Key Scientific Challenges award from Yahoo!, and multiple best-of-conference citations. He is an associate editor of SIGMOD Record and serves on the steering committee of the HILDA (Human-in-the-loop Data Analytics) Workshop. His research group is supported with funding from the NSF, the NIH, Adobe, the Siebel Energy Institute, and Google.

 

December 7th

Saul Gorn Lecture Series
Shafi Goldwasser
Electrical Engineering and Computer Science
MIT
"Pseudo Deterministic Algorithms and Proofs"

Read Abstract and Bio

Abstract

Probabilistic algorithms for both decision and search problems can offer significant complexity improvements over deterministic algorithms. One major difference, however, is that they may output different solutions for different choices of randomness. This makes correctness amplification impossible for search algorithms and is less than desirable in setting where uniqueness of output is important such as generation of system wide cryptographic parameters or distributed setting where different sources of randomness are used. Pseudo-deterministic algorithms are a class of randomized search algorithms, which output a unique answer with high probability. Intuitively, they are indistinguishable from deterministic algorithms by an polynomial time observer of their input/output behavior. In this talk I will describe what is known about pseudo-deterministic algorithms in the sequential, sub-linear and parallel setting. We will also briefly describe an extension of pseudo-deterministic algorithms to interactive proofs for search problems where the veri fier is guaranteed with high probability to output the same output on different executions, regardless of the prover strategies. Based on Joint work with Gat, Goldreich, Ron, Grossman and Holden.

Bio
Shafi Goldwasser is the RSA Professor of Electrical Engineering and Computer Science at MIT. She is also a professor of computer science and applied mathematics at the Weizmann Institute of Science in Israel. Goldwasser pioneering contributions include the introduction of interactive proofs, zero knowledge protocols, hardness of approximation proofs for combinatorial problems, and multi-party secure protocols.She was the recipient of the ACM Turing Award for 2012, the Gödel Prize in 1993 and another in 2001, the ACM Grace Murray Hopper award, the RSA award in mathematics, the ACM Athena award for women in computer science, the Benjamin Franklin Medal, and the IEEE Emanuel R. Piore award. She is a member of the AAAS, NAS and NAE.

Goldwasser received a BS degree in applied mathematics from Carnegie Mellon University in 1979, and MS and PhD degrees in computer science from the University of California, Berkeley, in 1984.