Wynne HSU

Provost's Chair Professor
Deputy Director, Institute of Data Science, NUS
Ph.D. (Electrical & Computer Engineering, Purdue University, U.S.A)
M.Sc. (Computer Science, Purdue University, U.S.A)
B.Sc. (Computer Science, National University of Singapore)
651 67944


Research Areas

  • Artificial Intelligence
  • Database
  • Computational Biology

Research Interests

  • Social Media Analytics
  • Spatio-temporal Data Mining
  • Retinal Image Analysis


Wynne Hsu received her BSc in Computer Science at National University of Singapore and her M.Sc. and Ph.D in Electrical Engineering from Purdue University, West Lafayette, U.S.A., in 1989 and 1994, respectively. She is currently a Provost's Chair Professor at the Department of Computer Science, School of Computing, National University of Singapore (NUS). Her research interests include: data analytics in the context of social networks, machine learning, as well as retina image analysis.

Current Projects

  • Retinal microvasculature as a “window” to study mechanisms and pathways in diabetic nephropathy (DN) - Diabetes leads to functional and structural alterations in the microvasculature in the kidney and retina. DN is a major microvascular complication and a significant cause of morbidity. Diabetic retinopathy (DR) often accompanies DN with clear parallels in their pathophysiology. In persons with diabetes, we and several others have shown early DR signs are associated with DN, end-stage kidney disease (ESKD)69 or worsening of kidney function. In addition to retinopathy changes, we have also shown retinal vascular imaging parameters measured from digital retinal images are associated with chronic kidney disease (CKD). In this project, we aim to define key features of microvascular abnormalities that predict DN. We propose to carry out this in two phases. In this first phase, we will enhance the cloud-based SIVA to allow for accurate registration of retina images and automatically quantified the changes in the retinal vascular structure. Further, to enable large-scale population studies, cloud-based SIVA has to be scalable. To this end, we intend to push more computations to high performance GPU processors. In the second phase, we will develop new predictive algorithms using the state-of-the-art “deep learning” techniques. These predictive algorithms take as input the quantified changes obtained in the first phase, as well as the clinical and laboratory data collected at baseline and subsequent follow-ups, and generate a set of base predictors. We will design a methodology that systematically determines the best ensemble of these base predictors in order to build the optimal predictor that can accurately determine the risk and progression of DN.
  • Assessing the Credibility of Information in New Media - As new media becomes an increasingly important form of communication and expression, novel analytic tools are needed to help society understand and better use this new media. Old media is comprised of curated forms of media, like newspapers, created and controlled by professionals who are trained to identify and correct misinformation. New media is not professionally edited or curated. Our research uses big data analysis tools to help understand the credibility of information in new media. We will develop novel, time-aware algorithms to help assess the credibility of such postings. We will study the flow of both credible information and misinformation to design strategies to diminish the influence of misinformation. The results of our work can be used by local industry to understand the public’s knowledge of their products or services (as expressed in social media), whether this knowledge is credible or not, and how to better propagate favorable information through a network. Our work could also be used by the Singapore government to assess public understanding (or misunderstanding) of important public health topics such as Dengue eradication.
  • Finding Supporting Opinions from Reviews - User generated content about products and services in the form of reviews are often diverse and even contradictory. This makes it difficult for users to know if an opinion in a review is prevalent or biased. We study the problem of searching for supporting opinions in the context of reviews.We propose a framework called SURF, that first identifies opinions expressed in a review, and then finds similar opinions from other reviews. We design a novel probabilistic graphical model that captures opinions as a combination of aspect, topic and sentiment dimensions, takes into account the preferences of individual authors, as well as the quality of the entity under review, and encodes the flow of thoughts in a review by constraining the aspect distribution dynamically among successive review segments. We derive a similarity measure that considers both lexical and semantic similarity to find supporting opinions. Experiments on TripAdvisor hotel reviews and Yelp restaurant reviews show that our model outperforms existing methods for modeling opinions, and the proposed framework is effective in finding supporting opinions.
  • Quantifying Bias in Ordinal Ratings - User opinions expressed in the form of ratings can influence an individual’s view of an item. However, the true quality of an item is often obfuscated by user biases, and it is not obvious from the observed ratings the importance different users place on different aspects of an item. We propose a probabilistic modeling of the observed aspect ratings toinfer (i) each user’s aspect bias and (ii) latent intrinsic quality of an item. We model multi-aspectratings as ordered discrete data and encode the dependency between different aspects by using a latent Gaussian structure. We handle the Gaussian-Categorical non-conjugacy using a stick-breaking formulation coupled with P´olya-Gamma auxiliary variable augmentation for a simple, fully Bayesian inference. We demonstrate the predictive ability of our model and its effectiveness in learning explainable user biases on two real world datasets.

Selected Publications

  • Wee Yong Lim, Mong Li Lee, Wynne Hsu. iFact: An Interactive Framework to Assess Claims from Tweets, in International Conference on Information and Knowledge Management (CIKM), Singapore, November 2017.

  • Lahari Poddar, Wynne Hsu, Mong Li Lee. Author-aware Aspect Topic Sentiment Model to Retrieve Supporting Opinions from Reviews, in International Conference on Empirical Methods in Natural Language Processing (EMNLP), Copenhagen, Denmark, September 2017.

  • Lahari Poddar, Wynne Hsu, Mong Li Lee. Quantifying Aspect Bias in Ordinal Ratings using a Bayesian Approach, in 26th International Joint Conference on Artificial Intelligence (IJCAI), Melbourne, Australia, August 2017.

  • Furong Li, Mong Li Lee, Wynne Hsu. Profiling Entities over Time in the Presence of Unreliable Sources, in IEEE Transactions on Knowledge and Data Engineering (TKDE), 2017.

  • Furong Li, Mong Li Lee, Wynne Hsu. MAROON+: A System for Profiling Entities over Time (Demo), in 33rd IEEE International Conference on Data Engineering (ICDE), San Diego, CA, USA, April 2017.

  • Chonggang Song, Wynne Hsu, Mong Li Lee. Temporal Influence Blocking: Minimizing the Effect of Misinformation in Social Networks, in 33rd IEEE International Conference on Data Engineering (ICDE), San Diego, CA, USA, April 2017.

  • Chonggang Song, Wynne Hsu, Mong Li Lee. Targeted Influence Maximization in Social Networks, in 25th ACM International Conference on Information and Knowledge Management (CIKM), Indianapolis, United States, October 2016.

  • Tien Yin Wong, Daniel Ting, Gavin Tan, Wynne Hsu, Mong Li Lee, Haslina Hamzah, Gilbert Lim, Rina Rudyanto, Yuan Cheng, Alfred Gan and Carol Y. Cheung. Cloud-based Automated Software for Diabetic Retinopathy Screening and Monitoring in a National Screening Program, in 39th Annual Macula Society Meeting, February 2016.

Awards & Honours

  • President’s Technology Award, Singapore, 2014.
  • SIGKDD Test-of-Time Award, 2014

Teaching (2020/2021)

  • CS6216: Advanced Topics in Machine Learning