The Road to Transparent AI: Latest Breakthroughs

Abstrat

In this keynote, I describe new breakthroughs on ML interpretability. This will include 1) powerful ways to interpret ML methods for time-series forecasting, clustering (phenotyping), and heterogeneous treatment effect estimation, 2) provide personalized explanations of ML methods which refer to the unique experience of the user of the ML method, and 3) autonomously discover scientific concepts using concept activation regions, which are generalizations of concept-based explanations. To learn more about our work in this area - see our website dedicated to this topic - https://www.vanderschaar-lab.com/interpretable-machine-learning/ and our github - https://github.com/vanderschaarlab/Interpretability