First Seminars Series

16:00 -17:00, Thursday, March to May 2021

Coordinator: SOH Yong Sheng

Neural Network Based Approximation Algorithm for Nonlinear PDEs with Application to Pricing

Ariel Neufeld
Nanyang Technological University, Singapore


Abstract  In this talk, we first recall the mathematical concept of neural networks and its universal approximation property. Then we will show how one can combine neural networks together with tools from Stochastic Calculus, particularly the Feynman-Kac representation, to build a neural network based algorithm that can approximately solve high-dimensional nonlinear PDEs in up to 10’000 dimensions with short run times. We apply this algorithm to price high-dimensional financial derivatives under default risk.

‏

Short Bio.
Ariel Neufeld obtained his PhD in mathematics from ETH Zurich under the supervision of Prof. Martin Schweizer (ETH Zurich) and Prof. Marcel Nutz (Columbia University), spending half of the time of his PhD at Columbia University in NYC. After his PhD he returned to ETH Zurich working as a postdoc in financial mathematics, insurance mathematics, and stochastic computation involving machine learning techniques. In 2019, he joined NTU Singapore having been awarded as Nanyang Assistant Professor in mathematics.

Healthcare Analytics: Learning from Multiple Heterogeneous Data Sources

Vaibhav Rajan
National University of Singapore, Singapore


Abstract  The increasing availability of digitized clinical and genomic data presents an unprecedented opportunity to study and gain deeper understanding of diseases, develop new treatments and improve healthcare ecosystems. However, clinical data also poses modelling challenges due to the heterogeneity of data sources, e.g., structured demographic variables, text in clinical notes, images in MRI scans etc.  In this talk, I’ll describe a deep multi-view learning technique, developed in my group, for unsupervised learning of representations from arbitrary collections of matrices. I’ll outline how our technique can be used for predicting gene-disease associations and drug targets in cancer.

‎

Short Bio. Vaibhav Rajan is an Assistant Professor in the Department of Information Systems and Analytics at the School of Computing, National University of Singapore (NUS). Earlier, he was a Senior Research Scientist at Xerox Research where he led a project on Clinical Decision Support Systems for over four years. He has also worked at Hewlett-Packard Labs and at Videoken, an education technology startup. Vaibhav Rajan received his PhD and Master’s degrees in Computer Science from the Swiss Federal Institute of Technology at Lausanne (EPFL), Switzerland in 2012 and 2008 respectively and his Bachelor’s degree in Computer Science from Birla Institute of Technology and Science (BITS), Pilani, India in 2004. His research interests include Machine Learning, Algorithm Design and their applications, primarily in Healthcare and Bioinformatics.

Information Newton’s flows: second-order optimization for Bayesian inference

Yifei Wang
Stanford University, USA


Abstract  We introduce a framework for Newton’s flows in probability space with information metrics, named information Newton’s flows. Extending the relationship between overdamped Langevin dynamics and Wasserstein gradient flows of Kullback-Leibler (KL) divergence, we derive Newton’s Langevin dynamics from Wasserstein Newton’s flows. We design sampling efficient variational methods in affine models and reproducing kernel Hilbert space (RKHS) to approximate Wasserstein Newton’s directions. Convergence results of the proposed information Newton’s method with approximated directions are established. Several numerical examples from Bayesian inference problems are shown to demonstrate the effectiveness of the proposed method.

‎

Short Bio. Mr. Yifei Wang is a PhD candidate at the Stanford University, USA.

Tensor Decomposition in Data Science

Joe Kileel
University of Texas at Austin, USA


Abstract  Tensors are higher-order matrices, and decomposing tensors can reveal structure in datasets.  In recent years, tensor decomposition has found applications in statistics, computational imaging, signal processing, and quantum chemistry.  In this talk, we will present a new numerical method for low-rank symmetric tensor decomposition, building on the usual power method and ideas from classical algebraic geometry.  The approach achieves a speed-up over the state-of-the-art by roughly one order of magnitude.  We will also describe an “implicit” variant of the algorithm for the case of moment tensors which avoids the explicit formation of higher-order moments, analogously to matrix-free techniques in linear algebra.  Time permitting, we will include a quantitative analysis of the non-convex optimization landscape associated with our algorithm.  This is based on joint works with Joao Pereira, Tammy Kolda and Timo Klock.

‎

Short Bio. Joe Kileel is an Assistant Professor at the Department of Mathematics and the Oden Institute of Computational Engineering and Sciences at the University of Texas at Austin, since Fall 2020.  Prior to this, he was a Simons postdoctoral fellow at the Program in Applied and Computational Mathematics, Princeton University.  He obtained a PhD in Mathematics from UC Berkeley in 2017, where his thesis was awarded the Bernard Friedman Memorial Prize in Applied Mathematics.  Joe’s research interests are in mathematics of data, computational algebra, and inverse problems.

When AI Meets Game Theory

Bo AN
Nanyang Technological University, Singapore


Abstract  In January 2017 CMU’s Libratus system beat a team of four top-10 headsup no-limit specialist professionals, which was the first time an AI had beaten top human players in this game. Libratus’s success is purely based on algorithms for solving large scale games and has nothing to do with deep learning! Over the last few years, algorithms for solving large scale games have also been applied to many domains such as security, sustainability, ad-word auction, and e-commerce. For some complex domains with strategic interaction, reinforcement learning is also used to learn an efficient policy. This talk will discuss key techniques behind these success and their applications in domains including games, security, e-commerce, and urban planning.

‏

Short Bio. Bo An is a President’s Council Chair Associate Professor in Computer Science and Engineering, and Co-Director of Artificial Intelligence Research Institute (AI.R) at Nanyang Technological University, Singapore. He received the Ph.D degree in Computer Science from the University of Massachusetts, Amherst. His current research interests include artificial intelligence, multiagent systems, computational game theory, reinforcement learning,  and optimization. His research results have been successfully applied to many domains including infrastructure security and e-commerce. He has published over 100 referred papers at AAMAS, IJCAI, AAAI, ICAPS, KDD, UAI, EC, WWW, ICLR, NeurIPS, ICML, JAAMAS, AIJ and ACM/IEEE Transactions. 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, and 2018 Nanyang Research Award (Young Investigator).  His publications won the Best Innovative Application Paper Award at AAMAS’12, the Innovative Application Award at IAAI’16, and the best paper award at DAI’20. 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 invited to be an Advisory Committee member of IJCAI’18.  He is PC Co-Chair of AAMAS’20. He is a member of the editorial board of JAIR and is the Associate Editor of AIJ,  JAAMAS, IEEE Intelligent Systems,  and ACM TIST.  He was elected to the board of directors of IFAAMAS and senior member of AAAI. 

Social technologies and sentiment analysis on population responses to COVID-19

Raj Kumar GUPTA and Yinping YANG
Institute of High Performance Computing, A*Star, Singapore


Abstract  What are social technologies? Why they are useful in understanding people in the digital era? How do these technologies help to understand how the population respond to the pandemic throughout its development? This joint presentation will introduce A*STAR’s affective and social research and advanced emotion analysis technologies, and discuss how they are developed and put in good use in a range of practical industry applications, including the context of COVID-19. We will first review conceptual concepts surrounding human emotions, machine learning, natural language processing, and system architecture that lay the foundation of our social technologies. We will then share a few key learning points from our COVID-19 social media analytics studies in collaboration with communication scientists and public health practitioners. Among our first published findings at JMIR Public Health and Surveillance, we discovered that at a global scale, public emotions shifted strongly from fear to anger over the course of the pandemic, while sadness and joy also surfaced. Our presentation will be concluded with a few key takeaways and discussion for future research.


Short Bio. Raj Kumar Gupta is a Senior Scientist at the Social and Cognitive Computing Department of the Institute of High Performance Computing, A*STAR, Singapore. His research interests include natural language processing, sentiment and emotion analysis, and multi-modal analysis. He received his PhD in Computer Science from the Nanyang Technological University, Singapore. He has more than five years’ industry experience on visual analytics and embedded systems.

Yinping Yang is a Senior Scientist and Principal Investigator of the A*STAR Digital Emotions programme at Institute of High Performance Computing, A*STAR, Singapore. Her research interests cover intelligent negotiation systems, emotion recognition and sentiment analysis, and strategic foresight. She obtained her Ph.D. in Information Systems from the National University of Singapore. Besides research, she also advises tech start-ups in emotion analytics and e-negotiation technologies. More detailed bio at: https://www.a-star.edu.sg/idlabs/people/adjunct-investigators/yang-yinping

Spectral Laplace-Beltrami Wavelets and Geometric Convolutional Neural Network
for Signal Processing and Classification  

Anqi QIU
National University of Singapore, Singapore


Abstract  The Laplace-Beltrami operator is a generalization of the Euclidean representation of the Laplace operator to an arbitrary Riemannian manifold. It is a self-adjoint operator and its eigenfunctions form a complete set of real-valued orthonormal basis functions. In this talk, I will introduce spectral Laplace-Beltrami wavelets and its computational algorithm. I will then demonstrate its use for smoothing and classification of the data defined on smooth surfaces embedded in the 3-D Euclidean space. Furthermore, I will discuss that the spectral Laplace-Beltrami Wavelets can be used for the construction of geometric convolutional neural network (CNN) and then introduce a vertex-based geometric CNN algorithm for regular surfaces in which translation and downsampling on surfaces can be the same as those in the regular grid. I will show the use of this method for the prediction of Alzheimer’s Disease.

‎

Short Bio. Anqi Qiu is an Associate Professor at Department of Biomedical Engineering, National University of Singapore. More detailed bio can be found at  https://www.eng.nus.edu.sg/bme/staff/dr-qiu-anqi/

Exploiting High-order Semantic Structures for Heterogeneous Information Networks

Yuan FANG
Singapore Management University, Singapore


Abstract  Heterogeneous Information Networks (HIN) often model complex graph-structured data
with multi-typed relationships between different types of objects. The heterogeneity gives rise to the rich semantics on HINs, such as different user interactions on social networks, and different functional associations between proteins on a biological network. Metagraphs, which model recurring subgraph structures on HINs, emerge as a powerful tool to capture the rich semantics. Compared to their simpler variant metapaths, metagraphs are higher-order structures with stronger expressiveness and can be regarded as nonlinear functions of metapaths. Not surprisingly, metagraphs can be used to derive effective graph representations on HINs, to enable various downstream tasks such as node classification and link prediction. The efficacy of metagraph-based representations can be demonstrated in various domains including social, cyberphysical and biological networks.

‏

Short Bio. Fang Yuan is currently an assistant professor at the School of Computing and Information Systems, Singapore Management University. He obtained his PhD degree in Computer Science from University of Illinois at Urbana-Champaign, and worked as a scientist/PI at Institute for Infocomm Research under Agency for Science, Technology and Research (A*STAR). His research focuses on graph-based data mining and machine learning, social network analysis, recommendation systems and bioinformatics. He has published extensively with more than 30 publications at premier international conferences and journals, including TKDE, KDD, VLDB, ICML, SIGIR and AAAI and IJCAI. In particular, his work on efficient PageRank computation on graphs was selected into the collection of the Best Papers of VLDB 2013. He has also contributed to various conferences and journals as PC/SPC member, Associate Editor and reviewer.

Unifying Online Optimization with Inventory Contraints and Revenue Constraints

Niangjun Chen
Singapore University of Technology and Design, Singapore


Abstract  Many problems in data science and cloud computing can be cast as online optimization problems. In this talk, we study general online optimization problems with uncertainties, where the payoff or cost information is only known to the decision-maker up to some bounded uncertainty set. We develop an algorithmic framework, CRT, that can be applied to both online cost minimization problems with revenue constraints and online revenue maximization problems with inventory constraints. In both scenarios, we design online algorithms based on the CRT framework and prove their optimality in terms of competitive ratio. Furthermore, while prior works on online nonlinear optimization with inventory constraint require the revenue functions to be concave, we show that CRT-based algorithms have optimal competitive ratios in general online optimization settings with non-convex cost functions or non-concave revenue functions. We demonstrate our proposed algorithms’ practical performance via trace-driven simulations in a cloud computing application against prominent existing algorithms.


Short Bio. Niangjun Chen received his B.S. in Computer Science at the University of Cambridge, and his M.S and Ph.D. in Computer Science at the California Institute of Technology. His research interests include optimization, machine learning, game theory, and their applications to complex systems such as smart grids, data centers, and transport. After graduation in 2017, he joined the Institute of High Performance Computing as a research scientist working on the optimization of logistics and the modeling and simulation of the transport systems. Since September 2020, he is an Assistant Professor at the Singapore University of Technology and Design. He has a joint appointment at the Institute for High Performance Computing at the Agency for Science, Technology, and Research (A*STAR). 

Medical Image Segmentation with Transformer

Yong LIU
Institute of High Performance Computing, Singapore


Abstract  Transformer has been developed as one of most impactful deep neural network architecture for Natural Language Processing (NLP) tasks such as machine translation, question & answer, and reading comprehension. Its effectiveness has been demonstrated by the impressive performance of large language model such as GPT-3 to produce convincing stream of human-like text. This talk will cover the mechanism of transformer and how it has been used for computer vision tasks such as image classification, followed by our recent work on transformer-based medical image segmentation for multiple diseases. Compared with representative existing methods, the proposed method consistently achieved the highest segmentation accuracy, and exhibited good cross-domain generalization capabilities. 


Short Bio. Dr. Liu Yong is the Deputy Department Director, Computing & Intelligence Department at Institute of High Performance Computing (IHPC), A*STAR, Singapore. He also holds a position as Adjunct Assistant Professor at Duke-NUS Medical School, NUS and Adjunct Principal Investigator at Singapore Eye Research Institute (SERI). He is also Chair of IEEE Systems, Man and Cybernetics Society (Singapore Chapter). Being passionate on the potential of AI on healthcare, he has led multiple research projects in multimodal machine learning, medical imaging analysis, especially AI for healthcare. He is currently the Program Lead (AI) for a large research program in AI for digital ophthalmology. He and his team have been awarded multiple NMRC grants. He has published papers in top tier AI and medical journals and conferences including New England Journal of Medicine, Nature Medicine, Lancet Digital Health, Annals of Neurology, AAAI, MICCAI, IROS, ICDE, and IJCAI.