#adversarial-learning2 articles

2 articles

AIBullisharXiv – CS AI · 5d ago7/103

🧠

Model Predictive Adversarial Imitation Learning for Planning from Observation

Researchers have developed a new approach called Model Predictive Adversarial Imitation Learning that combines inverse reinforcement learning with model predictive control to enable AI agents to learn from incomplete human demonstrations. The method shows significant improvements in sample efficiency, generalization, and robustness compared to traditional imitation learning approaches.

AIBullisharXiv – CS AI · Feb 277/106

🧠

On Discovering Algorithms for Adversarial Imitation Learning

Researchers have developed DAIL (Discovered Adversarial Imitation Learning), the first meta-learned AI algorithm that uses LLM-guided evolutionary methods to automatically discover reward assignment functions for training AI agents. This breakthrough addresses stability issues in adversarial imitation learning and demonstrates superior performance compared to human-designed approaches across different environments.