Bio
He is an associate professor of Computer Science at Stanford University where I am a member of the InfoLab and the AI lab. I joined the department in September 2009.
He is also an investigator at Chan Zuckerberg Biohub, where I focus on developing new methods for analysis of biomedical data.
His general research area is applied machine learning and data science for large interconnected systems. Focuses on modeling complex, richly-labeled relational structures, graphs, and networks for systems at all scales, from interactions of proteins in a cell to interactions between humans in a society. Applications include commonsense reasoning, recommender systems, computational social science, and computational biology with an emphasis on drug discovery.
His research has won several awards including a Lagrange Prize, Microsoft Research Faculty Fellowship, the Alfred P. Sloan Fellowship, and numerous best paper and test of time awards. It has also been featured in popular press outlets such as the New York Times and the Wall Street Journal. I received my bachelor's degree in computer science from University of Ljubljana, Slovenia, PhD in machine learning from Carnegie Mellon University and postdoctoral training at Cornell University.
Talks
-
Abstract
Graph structure of Neural Networks
Keynote
02-10-2022 - 17:15-18:00
Neural networks are often represented as graphs of connections between neurons. However, despite their wide use, there is currently little understanding of the relationship between the graph structure of the neural network and its predictive performance. In this talk we systematically investigate how does the graph structure of neural networks affect their predictive performance. To this end, we develop a novel graph-based representation of neural networks called relational graph, where layers of neural network computation correspond to rounds of message exchange along the graph structure. Using this representation we show that: (1) a "sweet spot" of relational graphs leads to neural networks with significantly improved predictive performance; (2) neural network's performance is approximately a smooth function of the clustering coefficient and average path length of its relational graph; (3) our findings are consistent across many different tasks and datasets; (4) the sweet spot can be identified efficiently; (5) top-performing neural networks have graph structure surprisingly similar to those of real biological neural networks. Our work opens new directions for the design of neural architectures and the understanding on neural networks in general.
