NUS Computer Science Research Week 2024

Program Details

Wednesday, 10/1/2024, 10:00 – 11:20
New Advances in Cryptography for Anonymity - Stefano Tessaro

Many applications have legitimate reasons to require the disclosure of some form of user information, e.g., to enforce access control, to collect statistics, or to detect fraud. On the one hand, we would like to minimize what is actually disclosed to avoid the unnecessary collection of personally identifiable information and to limit user tracking. On the other hand, it often appears challenging to achieve this without affecting the application’s intended functionality.
Over the last few decades, cryptographers have developed a number of theoretical tools that, in principle, can help us resolve this tension. This talk will discuss how recent widespread interest in industry has led to new research aimed at transitioning these tools into practice by developing solutions that are efficient, scalable, and theoretically sound. I will survey this recent progress, including our own research work, and highlight a number of challenges and open problems in this space.

Bio: Stefano Tessaro is an Associate Professor in the Paul G. Allen School of Computer Science & Engineering at the University of Washington, where he holds the Paul G. Allen Development chair. He works primarily on cryptography, computer security and theoretical computer science. Earlier, he was an assistant professor at UC Santa Barbara, where he held the Glen and Susanne Culler chair. He received a PhD from ETH Zurich in 2010, and held postdoctoral appointments at UC San Diego and MIT. He received several awards for his work, including a Sloan Research Fellowship, an NSF CAREER award, and a Hellman Fellowship, as well as a best-paper award at EUROCRYPT 2017.

Wednesday, 10/1/2024, 13:00 – 14:20
Alignment of Machine Learning Models: From CNNs to LLMs - Hamed Hassani

My goal in this talk is to give an overview of how the machine learning community has thought about the trustworthiness of trained models. This topic has been studied from several perspectives including the adversarial setting which is the focus of this talk. I'll start with adversarial examples and then make a (natural) transition to the recently emerged area of AI alignment wherein the broad objective is to “align” the output text generated by LLMs with ethical and legal standards. I will cover my research in these areas during the last few years ranging from attacks and defenses to fundamental limits and new notions of robustness. The talk will be self-contained and no particular background on the alignment of machine learning models will be needed.

Bio: Hamed Hassani is currently an associate professor of the Electrical and Systems Engineering Department, the Computer and Information Systems Department, and the Department of Statistics and Data Science at the University of Pennsylvania. Prior to that, he was a research fellow at Simons Institute for the Theory of Computing (UC Berkeley) affiliated with the program of Foundations of Machine Learning, and a post-doctoral researcher at the Institute of Machine Learning at ETH Zurich. He received a Ph.D. degree in Computer and Communication Sciences from EPFL, Lausanne. He is the recipient of the 2014 IEEE Information Theory Society Thomas M. Cover Dissertation Award, 2015 IEEE International Symposium on Information Theory Student Paper Award, 2017 Simons- Berkeley Fellowship, 2018 NSF-CRII Research Initiative Award, 2020 Air Force Office of Scientific Research (AFOSR) Young Investigator Award, 2020 National Science Foundation (NSF) CAREER Award, 2020 Intel Rising Star award, the distinguished lecturer of the IEEE Information Society in 2022-23, and the 2023 IEEE Communications Society & Information theory Society Joint Paper Award. Moreover, he has recently been selected as the recipient of the 2023 IEEE Information Theory Society’s James L. Massey Research and Teaching Award for Young Scholars.

Wednesday, 10/1/2024, 15:00 – 16:20
Online Algorithms: Beyond the Worst Case - Anupam Gupta

Decision-making in the face of uncertainty is a central topic across computer science: we all face questions of how to make good decisions without precise knowledge of the future, and of the events still to occur. Online algorithms and competitive analysis give us a rigorous worst-case framework in which to reason about such algorithmic problems. In this talk I will tell you about approaches to give more nuanced guarantees that go beyond the worst-case setting of competitive analysis. I will discuss proposed models and approaches in the context of some specific problems: set cover, network design, and online resource allocation.

Bio: Anupam Gupta is a Professor of Computer Science Department in the Courant Institute of Mathematical Sciences at New York University. Until recently, he was at Carnegie Mellon University; he did his PhD work at UC Berkeley, and his post-doctoral work at Cornell and Lucent Bell Labs. Anupam's research interests are broadly in the design and analysis of algorithms, primarily in algorithms for optimization in the face of uncertainty, in approximation algorithms for NP-hard optimization problems, and in understanding the algorithmic properties of metric spaces. He is an ACM Fellow, and a recipient of the Alfred P. Sloan Research Fellowship, the NSF Career award, and CMU's Herb Simon Award for Teaching Excellence.

Wednesday, 10/1/2024, 16:30 – 17:50
From audio to audio+ machine learning: creating common representational spaces with vision and language - Juan Pablo Bello

Machine learning research for sound detection, classification and localization, has long been hindered by the paucity of labeled audio data. However, the last few years have seen significant progress stemming from the use of multimodal self-supervised learning techniques coordinating audio with vision and language, what I term audio+. In this talk I first discuss our work in audio+ self-supervised learning and how it has resulted in significant improvement in sound detection and classification. Then I explore the limitations of current approaches in encoding spatial and temporal information in audio, and their effect on sound localization, description and general scene understanding tasks. I analyze how these limitations relate to basic assumptions embedded in the data, model architectures and learning objectives we use, and end by speculating the potential impact of alternative data and approaches on the future of audio and audio+ machine learning.

Bio: Juan Pablo Bello is a Professor of Music Technology, Computer Science & Engineering, Electrical & Computer Engineering, and Urban Science at New York University. In 1998 he received a BEng in Electronics from the Universidad Simón Bolívar in Caracas, Venezuela, and in 2003 he earned a doctorate in Electronic Engineering at Queen Mary, University of London. Juan’s expertise is in digital signal processing, machine listening and music information retrieval, topics that he teaches and in which he has published more than 150 papers and articles in books, journals and conference proceedings. Since 2016, he is the director of the Music and Audio Research Lab (MARL), a multidisciplinary research center at the intersection of science, technology, music and sound. Between 2019-2022 He was also the director of the NYU Center for Urban Science and Progress (CUSP). A fellow of the IEEE and a Fulbright scholar, his work has been supported by public and private institutions in Venezuela, the UK, and the US, including Frontier and CAREER awards from the National Science Foundation.

Thursday, 11/1/2024, 10:00 - 11:20
On Watermarking Generative AI in Generative AI Era - Yu-Xiang Wang

In this talk, I will give a tutorial style-talk that covers the motivation, challenges and recent advances associated with the problem of watermarking generative AI. Specifically, I will highlight the urgent need for watermarking AI generated content and then discuss two of our recent work on this problem. For text, I will present a simple watermark that comes with guaranteed quality (nearly indistinguishable from original), correctness (Type I / II errors) and security (against arbitrary edits).
For images, I will talk about how any invisible image watermarks can be certifiably removed using modern generative AI tools, and highlight a few possible ways to get around this attack. References: 1.[ZALW23] https://arxiv.org/abs/2306.17439 2. [ZZWL23] https://arxiv.org/abs/2306.01953

Bio: Yu-Xiang Wang is the Eugene Aas Associate Professor of Computer Science at UCSB. He directs the Statistical Machine Learning lab and co-founded the UCSB Center for Responsible Machine Learning. He is also a Visiting Academic with Amazon Web Services’s AI research lab. Yu-Xiang received his PhD in 2017 from Carnegie Mellon University (CMU), and his BEng and MEng from the National University of Singapore in 2011 and 2013 respectively. Yu-Xiang’s research interests include statistical theory and methodology, differential privacy, reinforcement learning, online learning and deep learning. His work had been supported by an NSF CAREER Award, Amazon ML Research Award, Google Research Scholar Award, Adobe Data Science Research Award and had received paper awards from KDD'15, WSDM'16, AISTATS'19 and COLT'21.

Thursday, 11/1/2024, 13:00 – 14:20
Beyond the computer vision comfort zone - Jean Ponce

Spectacular progress has been achieved in computer vision in the past dozen years, in large part thanks to black-box deep learning models trained in a supervised manner on manually annotated data, sometimes unrelated to any real task. I propose instead to give back to accurate physical models of image formation their rightful place next to machine learning in the overall processing and interpretation pipeline, and will discuss applications to two real engineering and scientific tasks, namely super-resolution and high-dynamic range imaging from photographic bursts acquired by handheld smartphones, and exoplanet detection and characterization in direct imaging at high contrast. In this context, realistic synthetic data are easy to generate without any manual intervention, but real ground truth is typically missing. I will also discuss new approaches to video prediction where real data is readily available, and training can be achieved in a self-supervised manner using temporal consistency. I will conclude by discussing potential real applications to this admittedly somewhat artificial problem.

Bio: Jean Ponce is a Professor at Ecole Normale Supérieure - PSL, where he served as Director of the Computer Science Department from 2011 to 2017 and a Global Distinguished Professor at the Courant Institute of Mathematical Sciences and the Center for Data Science at New York University. He is also the co-founder and CEO of Enhance Lab, a startup that commercializes software for joint demosaicing, denoising, super-resolution and HDR imaging from raw photo bursts. Before joining ENS and NYU, Jean Ponce held positions at Inria, MIT, Stanford, and the University of Illinois at Urbana-Champaign, where he was a Full Professor until 2005. Jean Ponce is an IEEE and an ELLIS Fellow and was a Sr. member of the Institut Universitaire de France. He has served as Program and/or General Chair of all three top international Computer Vision Conferences, CVPR (1997 and 2000), ECCV (2008) and ICCV (2023), and as Sr. Editor-in-Chief of the International Journal of Computer Vision. He currently serves as Scientific Director of the PRAIRIE Interdisciplinary AI Research Institute in Paris. Jean Ponce is the recipient of two US patents, an ERC advanced grant, the 2016 and 2020 IEEE CVPR Longuet-Higgins prizes, and the 2019 ICMLtest-of-time award. He is the author of "Computer Vision: A Modern Approach", a textbook translated in Chinese, Japanese, and Russian.

Thursday, 11/1/2024, 15:00 - 16:20
Privacy and Security Challenges in Machine Learning - Olya Ohrimenko

Machine learning on personal and sensitive data raises privacy concerns and creates potential for inadvertent information leakage (e.g., extraction of one’s text messages or images from generative models). However, incorporating analysis of such data in decision making can benefit individuals and society at large (e.g., in healthcare and transportation). In order to strike a balance between these two conflicting objectives, one has to ensure that data analysis with strong confidentiality guarantees is deployed and securely implemented.
My talk will discuss challenges and opportunities in achieving this goal. I will first describe attacks against not only machine learning algorithms but also naïve implementations of algorithms with rigorous theoretical guarantees such as differential privacy. I will then discuss approaches to mitigate some of these attack vectors, including property-preserving data analysis. To this end, I will give an overview of our work on ensuring confidentiality of dataset properties that goes beyond traditional record-level privacy (e.g., focusing on protection of subpopulation information as compared to that of a single person).

Bio: Olya Ohrimenko is an Associate Professor at The University of Melbourne which she joined in 2020. Prior to that she was a Principal Researcher at Microsoft Research in Cambridge, UK, where she started as a Postdoctoral Researcher in 2014. Her research interests include privacy and integrity of machine learning algorithms, data analysis tools and cloud computing, including topics such as differential privacy, dataset confidentiality, verifiable and data-oblivious computation, trusted execution environments, side-channel attacks and mitigations. Recently Olya has worked with the Australian Bureau of Statistics, National Bank Australia and Microsoft. She has received solo and joint research grants from Facebook and Oracle and is currently a PI on an AUSMURI grant. See https://oohrimenko.github.io for more information.

Thursday, 11/1/2024, 16:30 – 17:50
Designing Software for Certification - Natarajan Shankar

The versatility and flexibility of software makes it an indispensable tool for building critical systems, but its inherent complexity opens up vulnerabilities that can compromise safety and security. Software failures due to design flaws and bugs can be costly. These flaws are extremely expensive to fix once the software has been deployed. Safety-critical software systems need assurance that the software operates safely and securely prior to deployment. Such systems must therefore be designed with rigorous claims supported by reliable, reproducible, and maintainable evidence. We motivate the need for constructing software hand-in-hand with an assurance argument backing the critical safety and security claims. We describe some technologies that we have been developing to assist with design for certification. Specifically, we outline the ``efficient argument'' approach to system design, the use of formal architectures as a foundation for efficient compositional arguments, ontic type analysis linking the requirements ontology to code-level representations, automatic code generation from high-level specifications, and the Evidential Tool Bus (ETB) architecture for integrating evidence-generating tools within a design workflow for building and maintaining assurance arguments. The talk presents joint work with members of the DesCert (Design for Certification) project team.

Bio: Dr. Natarajan Shankar is a Distinguished Senior Scientist and SRI Fellow at the SRI Computer Science Laboratory. He received a B.Tech. degree in Electrical Engineering from the Indian Institute of Technology, Madras, and Ph.D. in Computer Science from the University of Texas at Austin. He is the author of the book, "Metamathematics, Machines, and Godel's Proof" (Cambridge University Press) and the co-developer of a number of technologies including the PVS interactive proof assistant, the SAL model checker, and the Yices SMT solver. He is a co-recipient of the 2012 CAV Award and the recipient of the 2022 Herbrand Award.

Friday, 12/1/2024, 10:00 – 11:20
Cryptography in the Wild - Kenneth Paterson

In this talk I’ll discuss the analysis of cryptographic systems as they are found in the wild. I’ll reflect on how we conduct this kind of research, why we do it, and what we can learn from it about how developers use (and misuse) cryptography.

Bio: Kenneth Paterson is a Professor of Computer Science at ETH Zurich, where he leads the Applied Cryptography Group. He is also the current Head of Department. Prior to joining ETH, he was a Lecturer, Reader and then Professor at Royal Holloway, University of London (2001-2019). He was also an EPSRC Leadership Fellow (2010-2015). Kenny was Editor-in-Chief of the Journal of Cryptology from 2017 to 2020 and Program Chair for Eurocrypt 2011. He was made a Fellow of the IACR in 2017 for research and service contributions spanning theory and practice, and for improving the security of widely deployed protocols. He is co-founder of the Real World Cryptography series of conferences. His research has won best paper awards at ACM CCS 2016 and 2022, IEEE S&P 2022 and 2023, NDSS 2012, CHES 2018, and IMC 2018. In 2022, he was winner of the "Golden Owl" best teaching award for the Department of Computer Science at ETH Zurich.

Friday, 12/1/2024, 13:00 – 14:20
Proving the security of real-world cryptography and protocols - Jonathan Protzenko

Cryptography and secure protocols are omnipresent in today's computing environment. Together, they form the cornerstone of modern computer security, powering a wide array of components such as secure web browsing (TLS), or private messaging services such Signal or WhatsApp.
Cryptography is not only hard to get right, but the consequences of failure are also catastrophic. Recognizing this, both industry and research have worked together to apply formal methods, and specifically software verification, to establish the correctness of cryptographic components with mathematical certainty. And today, if you are running Firefox, or an up-to-date version of the Python programming language, you are most likely using verified cryptography.
This lecture will provide an in-depth tour of the field of formal verification, and specifically its application to real-world cryptography, including deployment into widely used software. I will provide background on verification ; the various properties of interest one might want to establish when it comes to secure components ; and how to go about proving those in practice, using concrete examples from past research. Near the end of the talk, I will outline future directions and where cryptographic verification might be headed.

Bio: Jonathan Protzenko is a Principal Researcher in the RiSE group at Microsoft Research Redmond. His research focuses on advancing the theory and practice of software verification, i.e. showing with mathematical certainty that a critical piece of code exhibits the intended behavior. This is important for the software industry (e.g. cryptography), but also for society at large (e.g. the law).
His joint work (with many wonderful collaborators!) has received the Internet Defense Prize, and his code made it into the Linux kernel, the Python programming language, and the Firefox web browser, among others. His research on verified cryptography was featured in Quanta Magazine and IEEE Computer Magazine; his research on computational law appeared in Communications of the ACM.

Friday, 12/1/2024, 15:00 – 16:20
Rethinking software engineering research and education in the light of digital humanism - Carlo Ghezzi

The world in which we live relies on digital technologies, and in particular on software, which operates and interacts with the physical world and humans. In the digital era, software engineers are the demiurges who are creating a new cyber-physical world, where humans, autonomous agents powered by AI, and physical entities live together in a new kind of society. Already in the late 1990's constitutionalist L. Lessig said that software is the law that governs the world and asked for reflection and action, because of the potential disruptive consequences. This is even more urgent today, due to to the phenomenal progress of AI and AI-generated software, which led to an increasing pervasiveness of software-enabled functions, with more and more intimate relation with humans and society. This raises the urgent need for re-thinking the way we do research, the competences and responsibilities of technologists who conceive and develop software, and the skills they should acquire through education. Rethinking should start by asking questions like: Should software engineers care about the human values involved while conceiving/developing new applications? About possible future uses and ethical implications? Can they do it by themselves? What kind of skills would they need? The talk mainly aims at setting the stage for opening a much needed and urgent discussion, which should involve software researchers and educators and has to be broad and open, especially to social science and humanities.

Bio: Carlo Ghezzi is an Emeritus Professor of Computer Science at Politecnico di Milano, Italy, where he is currently Chair of the Ethical Committee.
He is an ACM Fellow, IEEE Fellow, member of Academia Europaea, member of the Italian Academy of Sciences (Istituto Lombardo). He has been awarded the ACM SIGSOFT Outstanding Research Award, the ACM SIGSOFT Distinguished Service Award, and the IEEE TCSE Distinguished Education Award. He has been on the board of several international research programs and institutions in Europe, China, Japan, and the USA. He has been President of Informatics Europe, the association of computer science departments and research laboratories in Europe and neighboring areas.
Carlo Ghezzi has been Program Co-Chair and General chair of several prestigious conferences (including the two flagship conferences on Software Engineering, ICSE and ESEC) and member of the program committee of many international conferences.
He has been Editor in Chief of the ACM Trans. on Software Engineering and Methodology, Associate Editor of Communications of the ACM, IEEE Transactions on Software Engineering, Science of Computer Programming. His research has been focusing on software engineering and programming languages.
He co-authored over 200 papers and 11 books, and coordinated several national and international research projects. He was a recipient of a prestigious Advanced Grant from the European Research Council. He is currently a Steering Committee member of the Digital Humanism Initiative and has recently co-edited a widely circulating open-access book on digital humanism.

Computer Science Research Week 2024

January 10 to 12

Registration

The event is free of charge. However, please register to help with the organization (to cater for coffee and lunch).

Speakers

Program

Program Details

Organizers

Venue

All talks will be held in Multi-Purpose Hall in the NUS School of Computing, COM 3 building.