Thursday, March 4, 2021 - 4:15pm to 5:15pm
Event Calendar Category
ORC
Speaker Name
Tommi Jaakkola
Affiliation
MIT
Abstract
Molecules are interesting objects from the point of view of machine learning. The available datasets are big, small, and heterogenous. The properties of molecules depend on intricate features of associated 2D graph structures, as well as 3D features that may not already be implicitly captured by graph representations. From the machine learning perspective, I will describe our efforts in developing tools to predict properties of compounds and their combinations, as well as associated challenges. I will also discuss generative molecular modeling -- de novo realization of novel compounds with desirable characteristics. The talk does not assume background in molecules but assumes familiarity with machine learning.
Biography
Tommi S. Jaakkola is the Thomas Siebel Professor of Electrical Engineering and Computer Science at MIT’s Department of Electrical Engineering and Computer Science and the Institute for Data, Systems and Society (IDSS), as well as an investigator at the Computer Science and Artificial Intelligence Laboratory (CSAIL). He is a AAAI Fellow with many awards for his publications. His research covers many aspects of machine learning, statistical inference, and methods for designing principled, interpretable solutions to large scale estimation problems involving incomplete data sources. His applied research focuses on recommendation, retrieval, and inferential tasks, the design and optimization of molecules and reactions for drug development, and causal modeling especially in a strategic context.