Sofiia Kasaeva

By leveraging a structured Fisher-based hypergradient, this approach reduces the complexity of inverse reinforcement learning, achieving competitive policy performance without the heavy computational burden of traditional methods.