Meta Learning and Evolutionary Strategies

The podcast episode discusses the connection between meta learning and evolutionary strategies in the context of reinforcement learning. The guest, Robert Tiako-Lange, shares his focus on meta learning and its connections to evolution and evolutionary strategies. He highlights the importance of finding the right inductive biases and explores how meta learning can go beyond traditional RL algorithms. He also mentions the potential of combining meta learning with evolutionary strategies to optimize agent behavior. The discussion delves into the concept of learning not to learn and the exploration of heuristic behaviors. Robert emphasizes the need for adaptive learning algorithms that can adjust to varying time budgets. Overall, the episode provides insights into the exciting developments in meta learning and its relation to evolutionary strategies.