Asymmetric RL

Reinformed Dreamer: An Asymmetric World Model Efficiently Trained through Latent Guidance

Jun 16, 2026

Informed Asymmetric Actor-Critic: Leveraging Privileged Signals Beyond Full-State Access

We extend asymmetric actor-critic methods to leverage arbitrary state-dependent privileged signals and introduce criteria to identify which signals most effectively improve learning in partially observable environments.

Apr 30, 2026

Informed Asymmetric Actor-Critic: Theoretical Insights and Open Questions

We present an asymmetric actor-critic method for partially observable environments that leverages arbitrary privileged information, without requiring full-state access, while preserving unbiased policy gradient estimates.

Jul 17, 2025