Research map / Reinforcement learning / Visual and representation RL
FeUdal Networks for Hierarchical Reinforcement Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use FuN's hierarchical Manager-Worker architecture instead of an LSTM baseline | Atari games requiring long-term credit assignment · Ms. Pacman, Amidar, Gravitar, Enduro, Frostbite | episode score | Helped |
| Use FuN instead of an LSTM baseline | Atari games · Hero, Space Invaders | episode score | No clear effect |
| Use FuN instead of an LSTM baseline | Atari games · Seaquest, Breakout | episode score | Hurt |
| Use FuN instead of an LSTM baseline | Atari agents · Alien | episode score | Hurt |
| Use FuN instead of an LSTM baseline | visual memory tasks · DeepMind Lab: Water maze, T-maze, T-maze+, Non-match | reward | Helped |
| Use FuN instead of Option-Critic | Atari agents · Asterix, Ms. Pacman, Zaxxon | score: more than 20x improvement on Asterix; doubles it on Ms. Pacman; more than triples it on Zaxxon | Helped |
| Transfer a FuN policy trained with action repeat 4 and adjust temporal settings for no action repeat | FuN · Atari environments without action repeat | episode score | Helped |
| Train Manager goals from Worker gradients and omit intrinsic reward instead of using transition policy gradients (ablation) | FuN · Ablation experiments | performance | Hurt |
| Train Manager goals with standard policy gradients from Gaussian-sampled goals instead of transition policy gradients (ablation) | FuN · Ablation experiments | performance | Hurt |
| Use absolute goals instead of directional goals (ablation) | FuN · Ablation experiments | performance | Hurt |
| Train the Worker using intrinsic reward alone, removing environmental reward (ablation) | FuN · Ablation experiments | performance | Hurt |
| Replace the Manager's dilated LSTM with a vanilla LSTM (ablation) | FuN · Ablation experiments | performance | Hurt |
Closest papers by meaning
- Learning Multi-Level Hierarchies with Hindsight (ICLR 2019 · similarity 0.87)
- An Efficient Approach to Model-Based Hierarchical Reinforcement Learning (AAAI 2017 · similarity 0.87)
- Learning Hierarchical World Models with Adaptive Temporal Abstractions from Discrete Latent Dynamics (ICLR 2024 · similarity 0.87)
- Enhancing Exploration and Exploitation in Hierarchical Reinforcement Learning with Subgoal Graph Learning (AAAI 2026 · similarity 0.86)
- Emergent Hierarchical Reasoning in LLMs through Reinforcement Learning (ICLR 2026 · similarity 0.86)
- Hierarchical Reinforcement Learning with Targeted Causal Interventions (ICML 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.