"Machine learning in chemistry and beyond" (ChE-651) seminar by Geemi Wellawatte: "Model Agnostic Counterfactual Explanations for Molecular Property Predictions"
Event details
| Date | 22.11.2022 |
| Hour | 14:00 › 15:00 |
| Speaker | Geemi Wellawatte is a PhD student at the University of Rochester, NY, with Prof. Andrew White. She completed her BSc. in Computational Chemistry in 2017 at the University of Colombo Sri Lanka. Her research work focuses on developing computational models to solve chemistry problems, particularly in coarse-grained modeling, deep learning in chemistry, and explainable AI. |
| Location | Online |
| Category | Conferences - Seminars |
| Event Language | English |
One of the challenges with deep learning is lack of model interpretability. This is a significant drawback in the chemistry domain, as lack of knowledge "why a certain prediction was made" dissuades chemists to trust predictions from deep learning. We propose a method that can provide local explanations for arbitrary models with the use of molecular counterfactuals. These are sparse explanations composed of molecular structures. A counterfactual is an example as close to the original, but with a different outcome. Although relatively new to AI, counterfactual explanations are a mature topic in philosophy and mathematics. We use counterfactuals to answer, "what is the smallest change to the features that would alter the prediction". Our Molecular Model Agnostic Counterfactual Explanations (MMACE), method traverses a local chemical space around a given base molecule to identify counterfactuals. Further, we introduce an open-source software named “exmol” that implements the MMACE algorithm for generating counterfactual explanations.
Practical information
- General public
- Free
Organizer
- Andres Bran, Jeff Guo, Kevin Maik Jablonka, Philippe Schwaller, Puck van Gerwen
Contact
- Andres Bran, Jeff Guo, Kevin Maik Jablonka, Philippe Schwaller, Puck van Gerwen