Research map / Reinforcement learning / Visual and representation RL
Making Sense of Reinforcement Learning and Probabilistic Inference
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 soft Q-learning without an epistemic uncertainty bonus | RL as inference · One unknown action | BayesRegret as the number of distractor actions N grows | Hurt |
| Add a posterior-based optimism bonus and a scheduled inverse temperature | Soft Q-learning · One unknown action | BayesRegret: soft Q-learning grows linearly in N; K-learning (≤2.2) | Helped |
| Use K-learning instead of soft Q-learning | DeepSea · DeepSea MDPs | time to learn | Helped |
| Use K-learning instead of soft Q-learning | bsuite agents · bsuite exploration tasks | summary score | Helped |
| Use Bootstrapped DQN instead of K-learning | bsuite agents · bsuite Scale tasks | summary score | Helped |
| Use an ensemble of K-networks rather than a single K-network (ablation) | K-learning deep RL implementation · bsuite | performance | Helped |
| Use soft Q-learning instead of K-learning | bsuite agents · bsuite bandit and mnist tasks | summary score | Hurt |
Closest papers by meaning
- Reinforcement Learning with Lookahead Information (NeurIPS 2024 · similarity 0.90)
- Breaking the Performance Ceiling in Reinforcement Learning requires Inference Strategies (NeurIPS 2025 · similarity 0.90)
- Bayesian Exploration Networks (ICML 2024 · similarity 0.89)
- From Static Policies to Adaptive Priors in Offline Reinforcement Learning (NeurIPS 2026 · similarity 0.88)
- Pausing Policy Learning in Non-stationary Reinforcement Learning (ICML 2024 · similarity 0.88)
- Enabling Realtime Reinforcement Learning at Scale with Staggered Asynchronous Inference (ICLR 2025 · similarity 0.88)
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.