Adversarial Optimization Insights
The discussion highlights the challenges of training robots to navigate complex behaviors that may not align with human actions. Adversarial optimization techniques, akin to those used in generative models, show promise in improving the robustness of reward functions. However, the difficulty of exploring the vast array of possible behaviors remains a significant hurdle, emphasizing the need for innovative approaches that don't require exhaustive coverage of all scenarios.In this clip
From this podcast
The Robot Brains Podcast
S3 E2 Stanford Prof Chelsea Finn: How to build AI that can keep up with an always changing world
Related Questions