Abstract: Counterfactuals, an active area of research in machine learning explainability, are explanations that produce actionable steps to move a data point from one side of a decision boundary to another. These explanations have a clear use-case for several applications ranging from loan decisions to healthcare diagnosis, where they need to advise stake-holders about actions they can take to achieve a different outcome. Individuals not provided loans want steps they can take to achieve a loan, and similarly patients want to know how they can achieve a better diagnosis.
This presentation showcases FastCFE, an algorithm and feature that uses reinforcement learning to provide real-time counterfactual explanations. Our presentation is broken down as follows:
1. Overview of Counterfactuals and Reinforcement Learning (RL)
2. Deep distributed reinforcement learning using OpenAI Gym and Ray+Rllib
3. Benchmarks and Results
Bio: Karthik Rao is a machine learning engineer at Arthur AI (Monitoring, Performance, Explainability). He was previously an undergraduate at Harvard where he focused on big data systems for machine learning. He is passionate about designing and building novel machine learning solutions using state-of-the art frameworks.