Select a calendar:
Filter February Events by Event Type:
SUNMONTUEWEDTHUFRISAT
Conferences, Lectures, & Seminars
Events for February
-
Dr. Karol Hausman (Google) - Preventing the Next Bitter Lesson
Tue, Feb 07, 2023 @ 10:00 AM - 11:00 AM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Dr. Karol Hausman, Google
Talk Title: Preventing the Next Bitter Lesson
Series: Robotics and Autonomous Systems Seminar
Abstract: Richard Sutton's essay titled "The Bitter Lesson" remains one of the most insightful observations of the last 70 years of AI progress. Given the recent trends and developments in AI and robotics, it might be useful to predict what our next bitter lesson will be, and how we can prevent it today. In this talk, I will explore one potential answer to this question. Based on the resulting insights, I will discuss how leveraging foundation models in robotics can benefit both. I will show various examples of how low-level tasks can be combined with large language models to enable performing complex and temporally extended instructions that were not possible before.
Biography: Karol Hausman is a researcher at Google and an adjunct professor at Stanford. He spent 5 wonderful years at USC where he did his PhD and he's super excited to be back to share his recent findings in robot learning
Host: Gaurav Sukhatme
Location: Olin Hall of Engineering (OHE) - 100B
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
CS Colloquium: Nicole Immorlica (Microsoft Research New England) - Data, the Fundamental Particle of Interaction
Mon, Feb 13, 2023 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Nicole Immorlica, Microsoft Research New England
Talk Title: Data, the Fundamental Particle of Interaction
Series: CS Colloquium
Abstract: Most economic models of interaction assume that agents hold beliefs in the form of priors, or probability distributions over a state of the world, which guide their behavior. In this talk, I consider a model in which beliefs are built off data or anecdotes that are drawn from a distribution parameterized by the state of the world and study how this impacts outcomes. I first discuss a model where agents communicate by sharing anecdotes. This mode of communication results in higher noise and bias when agents have differing preferences, giving rise to informational homophily and polarization. The results have implications for content regulation in social networks. Next, I discuss a model where a principal selectively discloses anecdotes to facilitate social learning. Here I will show that an appropriate information structure, chosen ex ante, can incentivize exploration and thus avoid the herding problems common in such social learning settings. The results have implications for the selection of reviews in online recommendation systems.
Based on joint work with Nika Haghtalab, Brendan Lucier, Jieming Mao, Markus Mobius, Divya Mohan, Alex Slivkins, and Steven Wu.
This lecture satisfies requirements for CSCI 591: Research Colloquium
Biography: Nicole Immorlica is a senior principal researcher at Microsoft Research New England (MSR NE) where she leads the economics and computation group. She is also chair of SIGecom, the ACM Special Interest Group on Economics and Computation, which fosters world-class research in this interdisciplinary field through conferences, awards, and mentorship programs. She received her BS in 2000, MEng in 2001 and PhD in 2005 in theoretical computer science from MIT in Cambridge, MA. She joined MSR NE in 2012 after completing postdocs at Microsoft in Redmond, WA and Centruum vor Wiskunde en Informatics (CWI) in Amsterdam, Netherlands, and a professorship in computer science at Northwestern University. Nicole's research interest is in the design and operation of sociotechnical systems. Using tools and modeling concepts from both theoretical computer science and economics, Nicole hopes to explain, predict, and shape behavioral patterns in various online and offline systems, markets, and games. She is known for her work on social networks, matching markets, and mechanism design. She is the recipient of a number of fellowships and awards including the Sloan Fellowship, the Microsoft Faculty Fellowship and the NSF CAREER Award. She has been on several boards including SIGACT, the Game Theory Society, and INFORMS Auction and Market Design Section; is an associate editor of Operations Research and Games and Economic Behavior, and was program committee member and chair for several ACM, IEEE and INFORMS conferences in her area.
Host: David Kempe
Location: Ronald Tutor Hall of Engineering (RTH) - 115
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
Bridging Ethical Algorithms, Law, and Practice: Hiring and Beyond
Wed, Feb 15, 2023 @ 02:00 PM - 03:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Dr. Swati Gupta, Assistant Professor, Stewart School of Industrial & Systems Engineering at Georgia Tech
Talk Title: Bridging Ethical Algorithms, Law, and Practice: Hiring and Beyond
Abstract: Optimization and statistical models based on historical and socio-economic data that do not incorporate fairness desiderata can lead to unfair, discriminatory, or biased outcomes. New ideas are needed to ensure that the developed systems are accountable under uncertainty and reduce a deeper propagation of biases in multi-level decisions. In this talk, I will first discuss methods for ensuring ethical hiring. Recent run-ins of Microsoft and Wells Fargo with the Labor Department's OFCCP highlight a paradox: failing to address workforce imbalance can result in legal sanctions and scrutiny, but proactive measures to address these issues might result in the same legal conflict. Dr. Gupta will propose that partially-ordered sets, "posets", can be used to transparently account for known uncertainties and biases in evaluation data, giving rise to an interesting class of optimization problems. We will showcase how to ensure a "competitive" online selection of candidates with this model. Keeping in mind the requirements of U.S. anti-discrimination law, however, certain methods can be construed as illegal (e.g., imposing quotas). I will discuss the tensions with the law and ways to argue the legal feasibility of our proposed approach. This is based on joint work with Deven Desai and Jad Salem. I will also briefly discuss ethical decision-making in the context of other applications such as admissions, detection of critical diseases like sepsis, facility location, and gerrymandering.
This lecture satisfies requirements for CSCI 591: Research Colloquium.
Biography: Dr. Swati Gupta is a Fouts Family Early Career Professor and Assistant Professor in the Stewart School of Industrial & Systems Engineering at Georgia Tech. She serves as the lead of Ethical AI in the NSF AI Institute on Advances in Optimization. She received a Ph.D. in Operations Research from MIT. Her research interests include optimization, machine learning, and algorithmic fairness, spanning various domains such as e-commerce, quantum optimization, and energy. She received the Class of 1934: Student Recognition of Excellence in Teaching in 2021 and 2022 at Georgia Tech, the JP Morgan Early Career Faculty Award in 2021, and the NSF CISE Research Initiation Initiative Award in 2019, and the Google Women in Engineering Award (India) in 2011. She was also awarded the prestigious Simons-Berkeley Research Fellowship in 2017-2018, where she was selected as the Microsoft Research Fellow in 2018. Her research and students have received recognition at various venues like INFORMS Doing Good with OR 2022 (finalist), MIP Poster 2022 (honorable mention), INFORMS Undergraduate Operations Research 2018 (honorable mention), INFORMS Computing Society 2016 (special recognition), and INFORMS Service Science Student Paper 2016 (finalist). Dr. Gupta's research is partially funded by the NSF and DARPA.
Host: USC Center for AI in Society
Location: Location: Register for the Zoom webinar here: https://usc.zoom.us/webinar/register/WN_kmziSvzGT0OBw-
Audiences: Everyone Is Invited
Contact: Asiroh Cham
-
CS Colloquium: Christophe Hauser (USC / ISI) - Binary program analysis for systems security: a journey of post-design security challenges
Tue, Feb 21, 2023 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Christophe Hauser, USC / ISI
Talk Title: Binary program analysis for systems security: a journey of post-design security challenges
Series: CS Colloquium
Abstract: Modern software stacks are complex and rapidly expanding. This continuous trend keeps raising new challenges for software security: the discrepancy between the number of trained human experts available and the growing scale of modern software makes traditional analysis techniques unfit to address security problems in a timely fashion in real-world settings. Existing solutions towards solving this conundrum are staggered across multiple stages in the software development process. While design-time approaches involving formal methods and proofs of correctness have received academic attention and demonstrated success in safety-critical domains such as aerospace, the current state-of-practice in most of the software industry relies on informal and reactive security techniques which often require manual analysis.
My work focuses on addressing the unique challenges of post-development security through principled approaches leveraging formal methods, reverse engineering and machine learning to detect, patch and prevent vulnerabilities across the software stacks. However, security properties are difficult to guarantee in the context of modern, real-world computer architectures and software engineering practices, and this difficulty is further exacerbated when source code, specification or design-level information is unavailable. Unfortunately, this context is very common when it comes to evaluating the security of third-party software, whether it is released in the form of applications, libraries or embedded firmware.
In this talk, I will present my research to date towards addressing these challenges by focusing on leveraging theoretically sound models while attempting to identify the best soundness trade-offs to make these practical and prioritize real-world impact.
More specifically, I will present applications of these models to the problems of vulnerability discovery in a post-development context, retrofitting security in binary code and on extending the scalability of vulnerability models with machine learning.
This lecture satisfies requirements for CSCI 591: Research Colloquium
Biography: Dr. Christophe Hauser is Research Computer Scientist and Research Lead at University of California's Information Sciences Institute, where he founded and co-leads the BASS (Binary Analysis and Systems Security) research group (https://urldefense.com/v3/__https://bass.isi.edu__;!!LIr3w8kk_Xxm!qKWHZjoxvzMpC-rGATAiOW1m9nqIFHGeItsBB8n2hqiYHcQ5pqEcPeMyuQgGrc1gg5tvklVajL8hTQ$ ).
His research focuses on multiple aspects of systems security including intrusion detection, vulnerability discovery, binary program analysis and reverse engineering. He has been publishing high-impact papers in top security conferences such as USENIX Security, the Annual Computer Security Applications Conference (ACSAC), USENIX Security, the Network and Distributed System Security (NDSS) Symposium and the IEEE symposium on Security and Privacy (S&P). He also has been actively serving as technical committee member for top security conferences, including as the ACM Conference on Computer and Communications Security (CCS), USENIX Security and ACSAC, and was part of the organizing committee of CCS 2022.
Prior to joining USC-ISI, he was a postdoctoral researcher in the Seclab at UC Santa Barbara where he worked on the design and development of the "angr" program analysis platform, which is now vastly used across academia and industry.
He received his Ph.D. degree in Computer Science from CentraleSupélec, University of Paris-Saclay, France, (jointly with Queensland university of technology, Australia).
Host: Department of Computer Science
Location: Olin Hall of Engineering (OHE) - 132
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
-
Can collaborative systems science methods improve intervention strategies to reduce alcohol-related problems?
Wed, Feb 22, 2023 @ 02:00 PM - 03:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Dr. Christina Mair, Associate Professor and Vice Chair for Research in the Department of Behavioral and Community Health Sciences at the University of Pittsburgh School of Public Health
Talk Title: Can collaborative systems science methods improve intervention strategies to reduce alcohol-related problems?
Abstract: Systems science approaches, such as agent-based models, provide the opportunity to compare a range of intervention implementation strategies in a simulated environment. The usefulness of a given model, however, is limited by its accuracy, focus, and salience to end users. In this talk, Dr. Mair will share a collaborator-designed systems model of alcohol-involved sexual violence on college campuses currently being developed through a series of collaborative model building sessions with a learning collaborative. Integrating collaborative model building with agent-based model development is an innovative, empirically-based approach that can improve implementation of effective strategies to address alcohol-involved sexual violence, promote preventive interventions, and stimulate campus-level policy and programmatic changes to reduce sexual violence among students.
This lecture satisfies requirements for CSCI 591: Research Colloquium.
Location: Register for the Zoom webinar here: https://usc.zoom.us/webinar/register/WN_O3DV23rCQNug1Plxj01MbA
Biography: Dr. Mair is an Associate Professor and Vice Chair for Research in the Department of Behavioral and Community Health Sciences at the University of Pittsburgh School of Public Health, with secondary appointments in the Clinical and Translational Sciences Institute and Department of Epidemiology. She also serves as Director of the Center for Social Dynamics and Community Health and Associate Director of the Public Health Dynamics Lab. Her research seeks to understand structural and contextual influences on substance use-related problems with the goal to reduce these problems in community settings.
Host: USC Center for AI in Society
More Info: https://cais.usc.edu/events/bridging-ethical-algorithms-law-and-practice-hiring-and-beyond/
Location: https://usc.zoom.us/webinar/register/WN_O3DV23rCQNug1Plxj01MbA
Audiences: Everyone Is Invited
Contact: Asiroh Cham
Event Link: https://cais.usc.edu/events/bridging-ethical-algorithms-law-and-practice-hiring-and-beyond/
-
CS Colloquium: Yang Liu (UC Santa Cruz) - Reliable Machine Learning: From Data to Deployment
Tue, Feb 28, 2023 @ 11:00 AM - 12:00 PM
Thomas Lord Department of Computer Science
Conferences, Lectures, & Seminars
Speaker: Yang Liu, UC Santa Cruz
Talk Title: Reliable Machine Learning: From Data to Deployment
Series: CS Colloquium
Abstract: Developing reliable machine learning systems presents challenges in handling biased input data and the consequences of deployment. For instance, a machine learning model for question answering (e.g., ChatGPT) can encode mistakes and biases that persist in the database; an unaware machine learning-powered decision-maker (e.g., for loan approval) can automatically deny people the chance of recourse, resulting in a decline of trust between human and machines; deploying a sequence of myopically optimized models may create an unfair "echo chamber" for users. The list goes on. This talk presents three challenges to building a reliable machine learning system: (1) developing fair and robust algorithms with biased training data, (2) auditing the dynamic interactions between users and machine learning models, and (3) maximizing the long-term welfare of machine learning ecosystems with efficient interventions. We will discuss our group's efforts in addressing these challenges.
This lecture satisfies requirements for CSCI 591: Research Colloquium
Biography: Yang Liu is an Assistant Professor of Computer Science and Engineering at UC Santa Cruz (2019 - present). He also leads the machine learning fairness team at ByteDance AI Lab. He was previously a postdoctoral fellow at Harvard University (2016 - 2018). In 2015, he received his Ph.D. degree from the Department of EECS at the University of Michigan, Ann Arbor. His research focuses on developing fair and robust machine learning algorithms to tackle the challenges of biased and shifting data. He is a recipient of the NSF CAREER Award. He has been selected to participate in several high-profile projects, including NSF-Amazon Fairness in AI, DARPA SCORE, and IARPA HFC. His research has observed deployments with FICO and Amazon. His recent work has been recognized with four best paper awards at relevant workshops.
Host: Vatsal Sharan
Location: Olin Hall of Engineering (OHE) - 132
Audiences: Everyone Is Invited
Contact: Assistant to CS chair
