Ameya Divekar

What to Visualize for Training Reinforcement Learning Agents

Visualization of reinforcement learning (RL) environment and learning dynamics of an agent is a vital step for debugging and better understanding of the learnt policy. Number of techniques will be presented to assist for debugging and monitoring convergence of an agent over complex domain. The problem of understanding agent learning is especially difficult when the state and action spaces are large and continuous. The talk with present insights gained from experiments to train such an agent on real world parameters data.

Leveraging reinforcement learning over continuous action spaces for autonomous system control. Leading Moonshots program for innovation bringing step change in value created at enterprise level using cutting edge AI technologies like NLP, Computer Vision and Deep Learning.

Expertise areas: Reinforcement Learning , GANs, Computer Vision, Natural Language Processing, Deploying ML Models, Amazon Web Services - S3, Lambda, EC2, Sagemaker, Azure ML

Patented technologies(applied) include : Staggered Pattern(CATIA), Semantic Painter (CATIA), Automatic Mate of Components using Machine Learning(Solidworks - filed), AI Driven Drawing Checker

This website uses cookies to ensure you get the best experience. Learn more