Professor Luke Ong is NTU's Vice President for AI and Digital Economy and the Dean for the College of Computing and Data Science. He joined NTU as a Distinguished University Professor in August 2022. Prior to joining NTU, he was Lecturer then Professor of Computer Science at the University of Oxford (1994-2022); Fellow of Merton College, Oxford (1994-2022); Honorary Professor of Computer Science, Bristol University (since 2022); Shaw Visiting Professor at National University of Singapore; and Prize Research Fellow, Trinity College, Cambridge (1988-1994). Professor Ong holds a B.A. in Mathematics (1984, Triple First), a Postgraduate Diploma in Computer Science (1985, Distinction) from Trinity College, University of Cambridge; and a PhD in Computer Science (1988) from Imperial College, University of London. 

Professor Ong's research is broad, ranging across semantics of computation, programming languages, verification, logic and algorithms, and algorithmic game theory. A notable feature of his work is the combination of ideas and methods from semantics and structures, with techniques from automated verification. Professor Ong is one of the leading figures and inventors of game semantics and its applications. His solution (with Hyland) to the PCF Full Abstraction Problem opened up the field of game semantics; and their constructions, known as Hyland-Ong games, have become standard notions in the semantics of programming languages. Professor Ong is also known for his pioneering contribution in the field of verification: his LICS 2006 paper co-initiated higher-order model checking, a new branch of algorithmic verification that combines ideas and methods from semantics with automata-theoretic and allied techniques in automatic verification, with applications to the verification of higher-order programs. His current research interests include computer and cyber security, higher-order logic and satisfiability modulo theories, and probabilistic and differentiable programming. Throughout his long stay in Oxford, Professor Ong has supervised well over 60 doctoral students and postdocs.

His contributions to the advancement of computer science have been recognised with leadership positions in major scientific conferences and bodies. Professor Ong was General Chair (2013-2015) of the ACM / IEEE Logic in Computer Science (LiCS), and Founding Vice Chair (2014-2019) of the ACM Special Interest Group in Logic and Computation. He was founding Steering Committee Chair (2015- 2018), Formal Structures for Computation and Deduction. He has served as programme chair and on the steering committees of leading scientific meetings, including ACM / IEEE LiCS, European Association of Theoretical Computer Science, European Association of Computer Science Logic, and European Joint Conferences on Theory and Practice of Software. Professor Ong has given numerous keynote presentations and invited lectures, including well over 100 at international research meetings.

Professor Ong has received several international and national accolades. He is the joint winner of the ACM / EATCS / EACSL / KGS Alonzo Church Award 2017 for Outstanding Contributions to Logic and Computation. He is also a recipient of the President of the Republic of Singapore Scholarship in 1981, Prime Minister’s Book Prize in 1980, Overseas Merit Scholarship from 1981 to 1984.

Talk Title
Neural Proofs for Sound Verification of Complex Systems

Abstract
I discuss the construction of sound proofs for the formal verification and control of complex stochastic models of dynamical systems and reactive programs.

Neural proofs are made up of two parts. Proof rules encode requirements for the verification of general temporal specifications over the models of interest. Certificates are then constructed from said proof rules with an inductive approach, namely accessing samples from the dynamics and training neural nets, whilst generalising such networks via SAT-modulo-theory queries, based on the full knowledge of the models.

In the context of sequential decision making problems over stochastic models, I discuss how to additionally generate policies/strategies/controllers, in order to formally attain given specifications.

Biography
Alessandro Abate is Professor of Verification and Control in the Department of Computer Science at the University of Oxford. Earlier, he did research at Stanford University and at SRI International, and was an Assistant Professor at the Delft Center for Systems and Control, TU Delft. He received a Laurea degree from the University of Padua and MS/PhD at UC Berkeley. His research work spans logics, probability, control and AI.

Talk Title
Agentic Reinforcement Learning

Abstract
LLM-based autonomous agents are playing an increasingly important role in many fields, including general computer control, software engineering, scientific discovery, and social simulation. Early implementations of autonomous agents largely relied on manually constructed workflows based on prompt engineering. However, such static workflows often struggle to generalize effectively to out-of-domain tasks, lack the ability to maintain high-quality, continuous interaction with environments and users, and are unable to achieve self-improvement during long-term operation. As a result, a growing body of research has begun to introduce reinforcement learning, enabling agents to continuously optimize their policies through interaction and thereby systematically improve their generalization, interaction, and tool-use capabilities. This talk will discuss major challenges encountered in agentic RL and some recent progress.

Biography
Bo An is a President’s Chair Professor at Nanyang Technological University, Singapore. He was elected a Fellow of the Association for the Advancement of Artificial Intelligence (AAAI). Dr. An was the recipient of the 2010 IFAAMAS Victor Lesser Distinguished Dissertation Award, an Operational Excellence Award from the Commander, First Coast Guard District of the United States, the 2012 INFORMS Daniel H. Wagner Prize for Excellence in Operations Research Practice, 2018 Nanyang Research Award (Young Investigator), and 2022 Nanyang Research Award. He was invited to give Early Career Spotlight talk at IJCAI’17. He led the team HogRider which won the 2017 Microsoft Collaborative AI Challenge. He was named to IEEE Intelligent Systems' "AI's 10 to Watch" list for 2018. He was PC Co-Chair of AAMAS’20 and General Co-Chair of AAMAS’23. He is on the IJCAI Board of Trustees and will be Program Chair of IJCAI’27. He was elected to the Executive Council of AAAI, the board of directors of IFAAMAS, and Distinguished member of ACM.

Talk Title
Testing and Analysis of Modern Software: Challenges and Opportunities

Abstract
Software development is undergoing significant transformations, with systems growing at an accelerated rate and increasingly incorporating AI-generated code.  As a result, successful software testing and analysis techniques such as fuzzing and symbolic execution have to similarly adapt to remain useful.  In this talk, I will discuss our recent research efforts in this direction, including testing and analysis approaches that leverage advances in AI while preserving many of the strengths of traditional methods.

Biography
Cristian Cadar is a Professor in the Department of Computing at Imperial College London, where he leads the Software Reliability Group (http://srg.doc.ic.ac.uk), working on automatic techniques for increasing the reliability and security of software systems.  Prof. Cadar's research has been recognised by several prestigious awards, including the EuroSys Jochen Liedtke Award, HVC Award, BCS Roger Needham Award, IEEE TCSE New Directions Award, Humboldt Research Award, and two test of time awards.  Many of the research techniques he co-authored have been open-sourced and used in both academia and industry.  In particular, he is co-author and maintainer of the KLEE symbolic execution system, a popular system with a large user base.  Prof. Cadar has a PhD in Computer Science from Stanford University, and undergraduate and Master's degrees from the Massachusetts Institute of Technology

Talk Title
Transforming Mobility: From Next-Visit Prediction to a Mobility Foundation Model

Abstract
Understanding where people move, when they move, and why they move is critical for applications ranging from transportation and urban planning to public health, safety, and disaster response. With the rapid growth of large-scale mobility data, a key challenge is how to build general-purpose models that can support many such applications, rather than being designed task by task.

In this talk, I will first introduce TrajGPT, our recent work on learning a transformer-based model of human mobility that can be trained at scale on unlabeled trajectory data and reused across multiple downstream tasks, such as completing missing mobility data, attributing visits to places, and detecting anomalous movement patterns—pointing toward the emergence of mobility foundation models.

More broadly, successful foundation models share three core properties: transferable representations of their basic units, self-supervised learning on massive unlabeled data, and a universal backbone adaptable to many tasks. TrajGPT demonstrates the latter two for mobility data. The remaining open challenge is identifying the right foundational units.

I will conclude by discussing geospatial objects (GEOs) as a promising direction for building richer, transferable representations that integrate mobility, environment, and spatial context—paving the way toward true mobility foundation models for urban computing, public health, and beyond.

Biography
Cyrus Shahabi is a Professor of Computer Science, Electrical & Computer Engineering and Spatial Sciences; Helen N. and Emmett H. Jones Professor of Engineering; and the director of the Integrated Media Systems Center (IMSC) at USC’s Viterbi School of Engineering.  He also served as USC's Thomas Lord Department of Computer Science from 2017 to 2022. He was co-founder of two startups, Geosemble Technologies and TallyGo, which both were acquired in July 2012 and March 2019, respectively. He received his B.S. in Computer Engineering from Sharif University of Technology in 1989 and then his M.S. and Ph.D. Degrees in Computer Science from the University of Southern California. He authored two books and more than three hundred research papers in databases, GIS, and multimedia, and he has over 14 US patents.

Dr. Shahabi has received funding from several agencies such as NSF, NIJ, NASA, NIH, DARPA, AFRL, IARPA, NGA, and DHS, as well as several industries such as Chevron, Cisco, Google, HP, Intel, Microsoft, NCR, NGC, and Oracle. He chaired the founding nomination committee of ACM SIGSPATIAL (2008-2011 term) and served as the chair of ACM SIGSPATIAL for the 2017-2020 term. He was an Associate Editor of IEEE Transactions on Parallel and Distributed Systems (TPDS) from 2004 to 2009, IEEE Transactions on Knowledge and Data Engineering (TKDE) from 2010 to 2013, VLDB Journal from 2009 to 2015 and PVLDB (Vol. 16) in 2023. He is on the ACM Transactions on Spatial Algorithms and Systems (TSAS) editorial board and ACM Computers in Entertainment. He was the founding chair of the IEEE NetDB workshop and the general co-chair of SSTD’15, ACM GIS 2007, 2008, and 2009. He has been PC co-chair of several conferences, such as APWeb+WAIM’2017, BigComp’2016, MDM’2016, DASFAA 2015, IEEE MDM 2013, IEEE BigData 2013 and VLDB 2024. He regularly serves on the program committee of major conferences such as VLDB, SIGMOD, IEEE ICDE, ACM SIGKDD, and IEEE ICDM.

Dr. Shahabi is a fellow of IEEE and NAI (National Academy of Inventors).  He received the ACM Distinguished Scientist Award 2009, the 2003 U.S. Presidential Early Career Awards for Scientists and Engineers (PECASE), the NSF CAREER award in 2002, and the 2001 Okawa Foundation Research Award. He received the ACM SIGSPATIAL 2023 10-Year Impact Award in 2023. He was also a recipient of the US Vietnam Education Foundation (VEF) faculty fellowship award in 2011 and 2012, an organizer of the 2011 National Academy of Engineering “Japan-America Frontiers of Engineering” program, an invited speaker in the 2010 National Research Council (of the National Academies) Committee on New Research Directions for the National Geospatial-Intelligence Agency, and a participant in the 2005 National Academy of Engineering “Frontiers of Engineering” program.

Talk Title
Algorithms and Time Complexity: Recent Advances

Abstract
This talk will discuss recent theoretical advances in the design of efficient algorithms. We will cover a range of results, including fast graph and string algorithms. The material is self-contained and structured to require only minimal prior knowledge of theoretical computer science.

Biography
Danupon Nanongkai is a Scientific Director at the Max Planck Institute for Informatics in Saarbruecken, Germany. His research focuses on graph algorithms and complexity theory, with a current emphasis on developing algorithmic techniques that are effective across a range of computational models. Honors he received include the  Principles of Distributed Computing Doctoral Dissertation Award (2013), an ERC Starting Grant (2016), and the Best Paper Award at FOCS (2022). 

Talk Title
Many Facets of Proportionality in Multiwinner Voting

Abstract
In a multiwinner election, voters express their preferences over candidates, and the goal is to select a fixed-size set of winners; these can be members of parliament, food items to be ordered for a reception, or statements about a controversial topic. This talk will summarize recent progress on defining what it means for an outcome of multiwinner election to be proportional, and how to extend this notion of proportionality to richer environments.

Biography
Edith Elkind is a Ginny Rometty Professor of Computer Science at Northwestern University. She obtained her PhD from Princeton in 2005, and worked in Israel, Singapore, and the UK before joining Northwestern in 2024. She works in algorithmic game theory, with a focus on algorithms for collective decision making. She is a recipient of the SIGAI Autonomous Agents Research Award and a Fellow of EurAI. She served as a chair of multiple leading conferences in AI and algorithmic game theory (including IJCAI, ACM EC, AAMAS, WINE and COMSOC), and serves as an editor in chief of Journal of AI Research.

Talk Title
From Human Intelligence to Machine Intelligence: a Brain-Inspired Agent Computation Architecture --- With A Case Study on Cybersecurity

Abstract
This talk investigates three important questions in agent research and development: how to build effective agents, how to endow them with advanced agentic capabilities—such as memory, knowledge digitization, reasoning, higher order thinking skills and general problem solving skills, and finally how to automate the construction and development of such agentic systems. To avoid the talk to be overly abstract, we use concrete applications in cybersecurity space to demonstrate how to automate cybersecurity expertise across the software development lifecycle, including vulnerability detection, diagnosis, proof-of-concept generation, and automated repair. The adoptability of this research could be applied to many other domains like vibe coding, medical analysis, material science and eventually auto-research. Finally, we discuss an interdisciplinary path toward Artificial General Intelligence (AGI), integrating insights from neuroscience, psychology, social sciences, and computer science to develop AI systems that are intelligent, agentic, and aligned with human values.

Biography
Dr. Liu Yang is currently a full professor in Nanyang Technological University, Executive Director of Cyber Security Research Centre @ NTU, and Executive Director of CyberSG R&D Programme Office (CRPO). In 2019, he received the University Leadership Forum Chair professorship at NTU, the President's Chair in 2024.

Dr. Liu specializes in software engineering, cybersecurity and artificial intelligence. His research has bridged the gap between the theory and practical usage of program analysis, data analysis and AI to evaluate the design and implementation of software for high assurance and security. Many of his research has been successfully commercialized. By now, he has more than 700 publications in top tier conferences and journals, and 30 best paper awards and one most influence system award in top software engineering conferences. He is also leading several major research centers and programs including Cysren, CRPO, Trustworthy AI in NTU (TAICeN) and CREATE center with ICL on medical device security. He has received a number of prestigious awards including MSRA Fellowship, TRF Fellowship, Nanyang Assistant Professor, Tan Chin Tuan Fellowship, Nanyang Research Award, ACM Distinguished Speaker, NRF Investigatorship and NTU Innovator (Entrepreneurship) Award.

Talk Title
A Camera-Centric Framework for Unified Multimodal Understanding and Generation

Abstract
Unified multimodal models promise to understand and generate in a single architecture, but most of today’s designs treat the camera in oversimplified ways, assuming fixed viewpoints or predefined fields of view. This limits their ability to handle real-world scenarios where perspectives shift and contexts are dynamic. In this talk, I will introduce Puffin, our new multimodal framework that brings the camera dimension into the picture. Puffin combines autoregressive and diffusion modeling to interpret and generate scenes from arbitrary viewpoints with continuous fields of view. A key idea behind Puffin is to treat the camera as language, allowing the model to “think with camera.” This aligns visual cues with photographic terms and grounds reasoning in physical context, making the model more spatially aware. Puffin is trained on a large dataset of four million vision–language–camera triplets, with both global camera parameters and pixel-level camera maps. This enables precise and flexible control over spatial generation. I’ll show how Puffin outperforms specialized baselines in controllable generation and camera understanding, and how, with instruction tuning, it generalizes to diverse tasks like spatial imagination, world exploration, and photography guidance.

Biography
Chen Change Loy is a President's Chair Professor with the College of Computing and Data Science, Nanyang Technological University, Singapore. He received his PhD (2010) in Computer Science from the Queen Mary University of London. Prior to joining NTU, he served as a Research Assistant Professor at the MMLab of the Chinese University of Hong Kong, from 2013 to 2018. His research interests include computer vision and deep learning with a focus on image/video restoration and enhancement, generative tasks, and representation learning. His accolades include the CCF-CV Test of Time Award, Nanyang Research Award, and IIT Bombay International Award. He served/serves as an Associate Editor of the Computer Vision and Image Understanding (CVIU), International Journal of Computer Vision (IJCV) and IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI). He also serves/served as the Area Chair of top conferences such as ICCV, CVPR, ECCV, ICLR and NeurIPS. He serves as the Program Co-Chair of CVPR 2026 and General Co-Chair of ACCV 2028.

Talk Title
An Agentic Approach to Optimize Agentic Workflows

Abstract
We are seeing an explosion in the use of autonomous agents in developing software. Such agents, often constructed using language models, now perform tasks from issuing queries to explore data stored in databases, to compiling and optimizing code. In this talk, I will first share some observations in studying workflows generated by agents for the natural language to SQL task, and then describe our current efforts towards using agents to tackle data-intensive programming tasks.

Biography
Alvin Cheung is an associate professor in the EECS department at UC Berkeley, where his group works on data management and programming language research. Work from his group has received a number of best paper / poster / demo awards in different venues. Alvin is a PECASE awardee (highest honor bestowed by the US government on early career scientists and engineers), a Sloan fellow, and a recipient of early career research awards from the data management research community, the programming languages research community, and various government agencies and companies.

Talk Title
Knowledge-informed Reinforcement Learning

Abstract
Domain knowledge from human expertise, physics, or safety constraints is inherently symbolic, whereas modern reinforcement learning (RL) methods typically use neural networks with distributed, approximate representations. This creates a mismatch between the types of knowledge we have (symbolic), which are discrete and discontinuous, and the representations into which we wish to incorporate them (neural), which are continuous, making integration within the RL framework challenging. This calls for neurosymbolic approaches that allow tight integration of symbolic and neural components.

This talk presents methods for incorporating such symbolic knowledge into RL, focusing on programmatic knowledge expressed as if–else rules with linear conditions (oblique decision trees). Here, we introduce DTSemNet, a neural encoding of symbolic programs that enables gradient-based training and straightforward integration into RL pipelines. Building on this representation, we explore learning barrier certificates to formally certify the safety of learned RL policies. Finally, we discuss a hierarchical RL framework that uses symbolic, abstract world models to support incremental knowledge acquisition and reasoning across possible belief worlds, improving planning and leading to better sample efficiency than conventional approaches that learn a full world model upfront.

Biography
Arvind Easwaran is an Associate Professor and Assistant Dean (Academic) in the College of Computing and Data Science at Nanyang Technological University (NTU), Singapore. He is also the co-Director for the CNRS@CREATE program DesCartes on AI-based robust decision-making for urban critical systems. He received a PhD degree in Computer and Information Science from the University of Pennsylvania, USA, in 2008. Prior to joining NTU in 2013, he has been an Invited Research Scientist at the Polytechnic Institute of Porto, Portugal, in 2009-2010, and a Research & Development Scientist at Honeywell Aerospace, USA, in 2010-2012. In NTU, his research focuses on the design and analysis of real-time and cyber-physical computing systems, including their application in domains such as automotive and urban energy systems.

Talk Title
Fair Division for Mixed Resources and Beyond

Abstract
The allocation of scarce resources among interested agents arises frequently in society, and we often want the chosen allocation to be fair to the participants, leading to what is known as the fair division problem. Despite its simple formulation, fair division contains rich mathematical and computational structure and has been a central topic in artificial intelligence and economics for many decades. In this talk, I will introduce basic models of fair division and give an overview of recent research directions, with the goal of giving the audience a sense of the different challenges that arise and the role computational thinking plays in addressing them. Among other things, I will discuss the challenge of defining fairness and designing algorithms when the resources to be allocated consist of a mix of divisible and indivisible goods.

Biography
Xiaohui Bei is an Associate Professor jointly appointed by SPMS and CCDS at Nanyang Technological University. He obtained his Ph.D. from Tsinghua University in 2012. His research interests include topics in resource allocation, computational economics, and general algorithm design. He has published more than 50 publications at top-tier computer science conferences and journals. He is the recipient of the Microsoft Research Asia Fellowship and the Nanyang Assistant Professorship.

Talk Title
High-Dimensional Vector Quantization: General Framework, Recent Advances, and Applications

Abstract
High-dimensional vector data lies at the core of numerous modern applications, from recommendation systems to large-scale retrieval and retrieval-augmented generation. Effectively managing and processing such data presents both significant opportunities and challenges. A key enabler in this context is vector quantization, which compresses high-dimensional vectors while preserving essential similarities. In this talk, I will begin by exploring why vector quantization is crucial for scalable and efficient high-dimensional vector management. I will then present the general framework of vector quantization and discuss existing popular schemes. Building on this foundation, I will introduce RaBitQ, a recent advance that provides optimized approaches for binary and scalar quantization and achieves asymptotic optimality. RaBitQ has been incorporated into multiple major production-level vector databases and search engines of Meta, ByteDance, Elastic, Apple, Alibaba, Milvus, OceanBase, etc. The talk will conclude with a discussion on applications of RaBitQ and future research directions.

Biography
Cheng LONG is an Associate Professor at the College of Computing and Data Science (CCDS), Nanyang Technological University (NTU). He earned his Ph.D. degree from Hong Kong University of Science and Technology (HKUST) in 2015. He has research interests broadly in data management and data mining. More recently, he has been working in high-dimensional vector data management (and its applications in large models such as retrieval-augmented generative AI). His work has garnered recognition and accolades, including the CCDS Research Award (Young Faculty), "Best Research Award" from ACM-Hong Kong, the "Fulbright-RGC Research Award" granted by the Research Grant Council (Hong Kong), and the "PG Paper Contest Award" bestowed by IEEE-HK.

Talk Title
Robust Sparsification via Sensitivity

Abstract
Robustness to outliers is important in machine learning. Many classical problems, including subspace embedding, clustering, and low-rank approximation, lack scalable, outlier-resilient algorithms. This paper considers machine learning problems of the form $\min_{x\in \mathbb{R}^d} F(x)$, where $F(x)=\sum_{i=1}^n F_i(x)$, and their robust counterparts $\min_{x\in\mathbb{R}^d} F^{(m)}(x)$, where $F^{(m)}(x)$ denotes the sum of all but the $m$ largest $F_i(x)$ values. We develop a general framework for constructing $\epsilon$-coresets for such robust problems, where an $\epsilon$-coreset is a weighted subset of functions $\{F_1(x),\dots,F_n(x)\}$ that provides a $(1+\eps)$-approximation to $F(x)$. Specifically, if the original problem $F$ has total sensitivity $T$ and admits a vanilla $\epsilon$-coreset of size $Q$, our algorithm constructs an $\epsilon$-coreset of size $O(\frac{mT}{\epsilon})+Q$ for the robust objective $F^{(m)}$. This coreset size can be shown to be near-tight for $\ell_2$ subspace embeddings. Our coreset algorithm has scalable running time and, by employing a sensitivity flattening argument, leads to new or improved algorithms for robust optimization problems, including regression and PCA. This talk is based on the joint work with Chansophea Wathanak In, David Woodruff and Xuan Wu.

Biography
Yi Li is an associate professor in the Division of Mathematical Sciences, with a joint appointment in College of Computing and Data Sciences, at Nanyang Technological University in Singapore. His main research interests lie in algorithms for massive datasets and sublinear time streaming algorithms, with a recent focus on randomized numerical linear algebra.

Talk Title
From Egocentric Perception to Embodied Intelligence: Building the World in First Person

Abstract
Egocentric perception offers a powerful foundation for building intelligent systems that understand and act in the world from a first-person perspective. In this talk, I present a unified research trajectory from perception to embodiment through three complementary works. EgoLM introduces a multimodal language model that grounds egocentric motion and video in natural language, enabling contextual understanding and generation of first-person actions. EgoLife extends this capability to long-horizon, real-world settings, demonstrating how continuous egocentric observations can support lifelong memory, reasoning, and assistance in daily human activities. Finally, EgoTwin explores generative embodiment by jointly synthesizing egocentric views and human motion, capturing the causal coupling between how an agent moves and what it sees. Together, these works illustrate a path toward embodied intelligence that builds, reasons about, and generates the world from within — not as an external observer, but as an experiencing agent.

Biography
Ziwei Liu is currently an Associate Professor at Nanyang Technological University, Singapore. His research revolves around computer vision, machine learning and computer graphics. He has published extensively on top-tier conferences and journals in relevant fields, including CVPR, ICCV, ECCV, NeurIPS, ICLR, SIGGRAPH, TPAMI, TOG and Nature - Machine Intelligence. He is the recipient of PAMI Mark Everingham Prize, MIT TR Innovators under 35 Asia Pacific, ICBS Frontiers of Science Award, CVPR Best Paper Award Candidate, Asian Young Scientist Fellowship and Singapore President's Young Scientist Award. He serves as an Area Chair of CVPR, ICCV, ECCV, NeurIPS and ICLR, as well as an Associate Editor of TPAMI and IJCV.

Talk Title
Beyond the Algorithm: The Surprising Reality of Fair ML Pipelines

Abstract
We often treat fairness in Machine Learning (ML) as a plug-and-play problem: pick a fair algorithm, run it on the data, and expect an unbiased result. But what happens when that algorithm is just one cog in a complex pipeline involving data cleaning, preparation, and hyperparameter tuning? This talk discusses some interesting findings, based on an extensive benchmarking of over 40 fair classification algorithms. These include for instance the equalizer effect, where proper data preparation alleviates the choice of a suited fair ML algorithm and insights on synergetic effects of combining algorithms. Results show that a more holistic approach to fair ML pipelines than the state of the art could hold the key to more robust models that balance fairness and accuracy.

Biography
Dr. Melanie Herschel is an Associate Professor at Nanyang Technological University, Singapore. Her research focuses on data integration, curation, provenance, and quality in complex data pipelines. With a strong background in both academic research and applied data systems tailored to various domains, she has led numerous projects at the intersection of scalable data processing and trustworthy data management. Her work has been published in leading conferences and journals in databases and information systems, and she actively contributes to shaping the future of data management through editorial and program committee roles.

Talk Title
Wireless Sensing in the IoT Era: Theories, Applications, and New Modalites

Abstract
Wireless technologies have achieved great success in data communication. In recent years, wireless signals (e.g., WiFi) have also been leveraged for sensing, enabling exciting applications such as passive localization, contact-free gesture recognition, and vital-sign monitoring. Despite promising progress, the potential of wireless sensing remains constrained by several fundamental challenges, including limited sensing range and poor robustness under device motion. In this talk, I will present our recent research advances: (i) on the theoretical side, enabling wireless sensing to operate reliably under device motion for the first time; (ii) on the application side, achieving fine-grained eye-blink detection using wireless signals; and (iii) on new sensing modalities, specifically quantum wireless sensing.

Biography
Jie Xiong is a Nanyang Associate Professor in the College of Computing and Data Science at Nanyang Technological University. He was previously a Principal Researcher at Microsoft Research Asia and an Associate Professor at the University of Massachusetts Amherst. Jie’s research interests include wireless sensing, mobile computing, and smart health. His work has received numerous prestigious awards, including the SIGMOBILE 2025 Rockstar Award, the SIGMOBILE 2024 Test-of-Time Award, the MobiCom 2024 Best Paper Award, Best Paper Runner-Up Awards at MobiCom 2022, 2021, and 2020, the SenSys 2022 Best Paper Award, the SECON 2022 Best Paper Award, the UbiComp (IMWUT) 2021 Distinguished Paper Award, and the CoNEXT 2014 Best Paper Award. He is also a recipient of the NSF CAREER Award and an NIH R01 Grant. Jie received his Ph.D. from University College London, his M.S. from Duke University, and his bachelor’s degree from Nanyang Technological University. His Ph.D. studies were supported by the Google European Doctoral Fellowship, and his dissertation was recognized as the Runner-Up for the 2016 British Computer Society Distinguished Dissertation Award. 

Talk Title
Expanding the Capabilities of the Web Platform with WebAssembly

Abstract
For over 20 years, JavaScript was the only language available for general-purpose programming on the Web. Now, WebAssembly has been introduced alongside JavaScript as an efficient compilation target for a variety of applications, allowing them to be deployed and executed to the Web with near-native performance. However, WebAssembly still suffers from a number of technical constraints, such as limited concurrency support, which prevent some applications from reaching their full potential on the Web. This talk will celebrate the existing successes of WebAssembly while highlighting current research and industrial experimentation which will further improve the language's expressiveness and performance.

Biography
Conrad Watt is a Nanyang Assistant Professor at Nanyang Technological University. Previously, he was a Research Fellow at Peterhouse, University of Cambridge. Conrad serves as a Chair of the W3C WebAssembly Community Group, the industrial standards body for the WebAssembly programming language. His research focusses on ensuring the correctness and security of WebAssembly and the wider Web ecosystem through formal verification, while developing new features for the language to enhance the expressiveness and performance of Web applications.

Talk Title
Proof Checking---The Last Mile in Trustworthy Automated Reasoning

Abstract
State-of-the-art automated reasoning tools are complex and highly-optimized pieces of software. This complexity can lead to an increased risk of soundness-critical bugs, which may affect the trustworthiness of automatically generated results. To remedy this state of affairs, many tools now generate proof logs (also called proof certificates), which can be independently checked for correctness. This talk is about the "last mile" in highly trustworthy automated reasoning---the development of efficient, formally verified proof checkers that are capable of soundly scrutinizing proof logs for various theories. I will discuss how our CakeML-based, formally verified proof checkers offer end-to-end verified correctness guarantees for the automated reasoning community. I will also present some ongoing work, where we argue that verification can enable bolder design of more complex proof systems while preserving trust in the overall certification process.

Related Webpage: https://cakeml.org/checkers.html

Biography
Yong Kiam TAN is a Nanyang Assistant Professor in the College of Computing and Data Science at Nanyang Technological University, Singapore. He also holds a joint appointment as a research scientist at the Institute for Infocomm Research, A*STAR, Singapore. He completed his PhD in Computer Science (Pure and Applied Logic) at Carnegie Mellon University, advised by Prof. André Platzer. He is interested in applications of deductive verification and interactive theorem proving in areas such as automated reasoning, compilers, formalized mathematics, hybrid systems, and security (cryptography).

Talk Title
Holistic Approach for Designing Intelligent and Resource-Efficient AI Inference Systems

Abstract
As AI inference workloads become a primary driver of cloud consumption, conventional infrastructure is challenged to provide the required scale and efficiency. In this talk, I will present a forward-looking perspective on building next-generation scalable resource-efficient AI inference systems, advocating for a holistic co-design of AI applications and the cloud stack itself. I will introduce this approach through two innovative systems: ServerlessLLM and Slim-SC. ServerlessLLM reimagines resource management to dramatically accelerate model serving, while Slim-SC introduces an optimized reasoning approach that prunes redundant computations without sacrificing accuracy. By examining the design of these systems, we can uncover a conceptual framework for developing more powerful, scalable, and cost-effective intelligent platforms. Finally, we will discuss a vision for the future of cloud computing for AI systems, where AI-native infrastructure seamlessly and efficiently powers the next generation of intelligent applications.

Biography
Dr. Dmitrii Ustiugov is an Assistant Professor at NTU Singapore. Previously, Dmitrii was a Postdoctoral Researcher at ETH Zurich after receiving a Ph.D. from the University of Edinburgh. Dmitrii’s research interests lie at the intersection of Computer Systems and Architecture, with a current focus on support for serverless cloud and large-scale AI systems. His works are published in top-tier computer systems, architecture, and ML venues, such as OSDI, ASPLOS, ISCA, EMNLP. Dmitrii’s work was recognized by MIT TechReview Asia-Pacific 2024 Innovators under 35 (Visionary), ASPLOS’21 Distinguished Artifact, and MICRO TopPicks Honorable Mention 2023.

Talk Title
Understanding Visual Intelligence Through Physical Intrinsics

Abstract
Much of our visual world has an intrinsic, physical structure: scenes are composed of objects; objects possess their own geometry, texture, material, and physical properties. How can we infer, represent, and use such intrinsic structure from raw visual data, without hampering the expressiveness of neural networks? Alternatively, with the rapid development of visual AI models, what role does such structural information play, or do we still need it at all? In this talk, I will discuss our recent efforts on machine visual understanding, reconstruction, and generation, and their connections to such physical intrinsics. I will introduce and contrast two technical paths: leveraging intrinsics as powerful inductive biases vs. grounding pre-trained vision foundation models onto intrinsics. I will show that we can now build visual intelligence that infers object shape, texture, material, and physics, as well as scene context, all from a single image or video, with applications in controllable, action-conditioned 4D visual world understanding, generation, and interaction.

Biography
Jiajun Wu is an Assistant Professor of Computer Science and, by courtesy, of Psychology at Stanford University, working on computer vision, machine learning, robotics, and computational cognitive science. Before joining Stanford, he was a Visiting Faculty Researcher at Google Research. He received his PhD in Electrical Engineering and Computer Science from the Massachusetts Institute of Technology. Wu's research has been recognized through the Young Investigator Programs (YIP) by ONR and by AFOSR, the NSF CAREER award, the Okawa research grant, the AI's 10 to Watch by IEEE Intelligent Systems, paper awards and finalists at ICCV, CVPR, SIGGRAPH Asia, ICRA, CoRL, and IROS, dissertation awards from ACM, AAAI, and MIT, the 2020 Samsung AI Researcher of the Year, and faculty research awards from Google, J.P. Morgan, Samsung, Amazon, and Meta.

Talk Title

Past, Present and Future Challenges in Datacenter and Cloud Networking

Abstract

Datacenter and cloud networks are undergoing a fundamental transition. Up until recently, datacenter networking has been driven by increasing link speeds, shrinking switch buffers, tight latency requirements, and in-network programmability. While such bespoke solutions are still hard to deploy in a multi-tenant cloud environment, hence are not widely accessible to cloud tenants, emerging AI workloads and agentic applications are shaking the requirements for datacenter and cloud networking by changing the traffic patterns and introducing new communication protocols.

In this talk, I will first present SIRD (NSDI’25), corresponding to the past challenges, a congestion control protocol designed for modern datacenter fabrics. SIRD revisits receiver-driven designs and shows how explicitly distinguishing between single-owner and shared links enables precise scheduling without sacrificing stability. By combining scheduling with reactive control, SIRD achieves high utilization and near-optimal latency while keeping queuing minimal, even at 100 Gbps.

I then, turn to current challenges that revolve around making all this research on datacenter infrastructure available to the public cloud. I present KRAKENGUARD, a policy-driven access control framework that enables safe, multi-tenant use of eBPF specifically for networking hooks such as XDP. KRAKENGUARD enforces fine-grained constraints on eBPF programs at load time based on exhaustive symbolic execution, preventing privilege abuse and unsafe interference between co-located network functions.

Finally, I will describe our ongoing effort (future challenges) towards describing a networking stack specifically targeting AI training workloads. I will explain how the specifics of the AI training communication patterns allow us to completely rethink the required mechanisms for routing, congestion control, and Quality of Service.

Biography

Marios Kogias is an Assistant Professor in the Department of Computing at Imperial College London, where he conducts research in operating systems, networking, and distributed systems, with a particular focus on tail-tolerant systems, datacenter networking, and cloud infrastructure. He received his PhD from EPFL, where his work was recognised with the 2021 Dennis M. Ritchie Doctoral Dissertation Award and an honourable mention for the Roger Needham PhD Award, and was supported by an IBM PhD Fellowship. His research has been published in top-tier systems conferences and has received a Best Student Paper Award at Eurosys and a Distinguished Artifact Award at ASPLOS. He is currently supported by an ERC Starting Grant, EPSRC funding, and industry collaborations

Talk Title
Algebraic Methods in Algorithms and Complexity

Abstract
Efficient computation lies at the heart of modern computer science. A central goal of complexity theory is to understand the resources—time, space, and randomness—needed to solve computational problems across different computational models. This talk focuses on algebraic complexity, where the objects are (multivariate) polynomials . Polynomials are ubiquitous: they encode counting and optimization problems, capture structure in pseudorandomness, coding theory, and probability theory. Often they provide a bridge between algorithms and lower bounds. I will briefly talk about a few current directions, including circuit lower bounds, derandomization, and algebraic approximations for complexity measures, highlighting connections to Geometric Complexity Theory (GCT), a major program toward proving the famous P vs NP conjecture.

Biography
Pranjal Dutta is an Assistant Professor in the College of Computing and Data Science (CCDS) at Nanyang Technological University (NTU), Singapore. Previously, he was a Jane Street Fellow and a Simons–Berkeley Research Fellow (Fall 2025). Before that, he was a postdoctoral researcher at the National University of Singapore with Divesh Aggarwal. He received his PhD from the Chennai Mathematical Institute (CMI), advised by Nitin Saxena (IIT Kanpur), where his thesis won the ACM India Best Dissertation Award in 2023. He was also a Google PhD Fellow. His research spans complexity theory, coding theory and pseudorandomness.

Talk Title
Test Before You Trust: Verifying Predictions in Online Allocation and learning

Abstract
Learning-augmented algorithms aim to combine predictive advice with worst-case guarantees: they achieve near-optimal performance when predictions are accurate, while remaining robust when predictions are unreliable. A central challenge, however, is determining when predictions should actually be trusted—especially when their quality is unknown a priori.

In this talk, I present a principled “test-before-trust” paradigm for both online allocation and distribution learning. The key idea is to statistically validate predictions using a small testing phase before committing to prediction-driven decisions. Leveraging tools from distribution property testing, we design algorithms whose performance smoothly interpolates between classical worst-case guarantees and improved bounds when advice is accurate.

I will illustrate this framework in three settings. In online bipartite matching with imperfect advice, we establish tight impossibility results under adversarial arrivals and give testing-based algorithms under random order that adapt to advice quality. In high-dimensional Gaussian learning with predicted parameters, we show that advice can reduce the classical Õ(d²/ε²) sample complexity when sufficiently accurate, while retaining worst-case guarantees otherwise. Finally, for product distributions over {0,1}^d, we prove that although Ω(d/ε²) samples are necessary in the worst case, advice that is close in ℓ₁-distance enables learning with Õ(d^{1−η}/ε²) samples—without knowing the advice error in advance.

Together, these results demonstrate that statistical validation provides a unifying mechanism for controlling the robustness–consistency tradeoff across online allocation and high-dimensional learning.

Biography
Themis Gouleakis is an Assistant Professor in the College of Computing and Data Science at Nanyang Technological University. Prior to this, he was a postdoctoral researcher at the University of Southern California, supervised by Ilias Diakonikolas, and a postdoctoral fellow in the Algorithms and Complexity Department at the Max Planck Institute for Informatics. He also served as a Senior Research Fellow at the National University of Singapore, where he was supervised by Arnab Bhattacharyya and Vincent Y. F. Tan.

He received his Ph.D. from MIT, where he was affiliated with the Theory of Computation group in CSAIL and advised by Ronitt Rubinfeld.

He received the Outstanding Paper Award at the 33rd Annual Conference on Neural Information Processing Systems (NeurIPS 2019).

Talk Title
Breaking Free from POSIX: Rethinking Consistency Models for HPC File Systems

Abstract
POSIX has long served as the de facto standard for specifying and implementing file systems. However, it was originally designed (decades ago) for single-machine environments with a single storage device, not for the scale and complexity of today's HPC systems. Over the years, the strict semantics imposed by POSIX has become a significant bottleneck, hindering the performance and scalability of HPC file systems. This challenge has been further exacerbated by the rapid growth of HPC system scale, the advent of fast storage devices, and the increasing I/O demands of emerging workloads, such as AI training and high-resolution scientific simulations. In this talk, I will explore ongoing advances in HPC storage systems and address a fundamental question in the community: Do HPC applications truly need POSIX? I will delve into why the POSIX is inherently unsuitable for HPC and explore alternative solutions better aligned with HPC needs.

Biography
Chen Wang is an Assistant Professor at the College of Computing & Data Science, Nanyang Technological University. Prior to joining NTU, he was the Fernbach Postdoctoral Fellow at the Center for Applied Scientific Computing (CASC) at Lawrence Livermore National Laboratory. He received his Ph.D. in Computer Science at the University of Illinois Urbana-Champaign, under the guidance of Prof. Marc Snir. His research interests include High-performance Computing (HPC), Parallel Storage System, System for AI/LLM, and Performance Analysis. His work has been published in premier HPC and system venues, including SC, HPDC, TPDS, IPDPS, etc. He has an Erdős number of 3.