Prof. Limsoon Wong (黄任祥)

National University of Singapore, Singapore

Limsoon Wong is Kwan-Im-Thong-Hood-Cho-Temple Professor in the School of Computing and Honorary Professor of Pathology at the National University of Singapore. His research focuses on knowledge discovery in biomedicine. He is a Fellow of the ACM and the Singapore National Academy of Science.

Speech Title: "Rethinking benchmarking beyond the leaderboard"

Abstract: Benchmarking has become indispensable for evaluating methods in computational biology. But are our benchmarks really telling us what algorithms can do? Most benchmark studies reduce performance to aggregate scores that facilitate ranking while concealing important differences in capability, the conditions under which methods succeed, and the ways in which they fail. We argue that benchmarking should be treated as scientific measurement. Its purpose is not simply to identify a winner, but to discover what an algorithm can and cannot do. This requires four principles for diagnostic benchmarking: designing informative assessment groups, choosing metrics that answer the right questions, applying explicit rules to interpret performance patterns, and reporting diagnostic profiles rather than relying on aggregate scores. We apply this perspective to several areas of computational biology, including single-cell RNA-seq integration, protein structure prediction, and protein function prediction. Along the way, we uncover surprising deficiencies in prevailing benchmarks and surprising weaknesses in methods that conventional scores can make look strong. The message is simple: a benchmark is not merely a leaderboard; it is an experiment on an algorithm.

 

Prof. Andrew Teschendorff

Shanghai Institute of Nutrition and Health, Chinese Academy of Sciences, China

Andrew Teschendorff is a Principal Investigator at the Shanghai Institute of Nutrition and Health, part of the Chinese Academy of Sciences, where he leads a lab in Computational Systems Epigenomics. His research interest is in elucidating the role that epigenetic changes play in aging and cancer-risk. He trained as a Mathematical Physicist at the University of Edinburgh (BSc, 1990-1995) and Cambridge University (MASt & PhD, 1996-2000). After a first postdoc in Mathematical Biology, he held Cambridge-MIT and Isaac Newton Fellowships in Statistical Cancer Genomics at the University of Cambridge (2003-2008). From 2008 to 2013 he held the Heller Research Fellowship at the UCL Cancer Institute in London. From 2015 to 2019 he held an Advanced International Newton Fellowship from the Royal Society in association with UCL London. In 2023 he received a Clarivate Highly Cited Researcher award in recognition of how his work has influenced the epigenomics, aging and cancer systems biology fields. He holds various patents on algorithms for cancer risk prediction and cell-type deconvolution, is an Associate Editor for Genome Biology, Genome Medicine and Clinical Epigenetics, and is a regular reviewer and statistical advisor for Nature, Science and other premier journals.