Research map / Reinforcement learning / Policy gradient and actor-critic
Distributional Policy Optimization: An Alternative Approach for Continuous Control
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 GAC AIQN instead of the strongest policy-gradient baseline | continuous-control agents · MuJoCo suite | relative result: Humanoid-v2: +3447 (+595%); Walker2d-v2: +533 (+14%); Hopper-v2: +467 (+17%); HalfCheetah-v2: -381 (-4%); Ant-v2: -444 ( | Mixed |
| replace autoregressive AIQN actor with non-autoregressive IQN (ablation) | GAC · Hopper-v2 | maximal attained value: 3234 ± 122 -> 1473 ± 421 | Hurt |
| replace autoregressive AIQN actor with non-autoregressive IQN (ablation) | GAC · Humanoid-v2 | maximal attained value: 4056 ± 878 -> 3547 ± 572 | Hurt |
| replace autoregressive AIQN actor with non-autoregressive IQN (ablation) | GAC · Walker2d-v2 | maximal attained value: 4357 ± 160 -> 1390 ± 651 | Hurt |
| replace autoregressive AIQN actor with non-autoregressive IQN (ablation) | GAC · Swimmer-v2 | maximal attained value: 238 ± 3 -> 45 ± 0 | Hurt |
| replace autoregressive AIQN actor with non-autoregressive IQN (ablation) | GAC · Ant-v2 | maximal attained value: 5064 ± 208 -> 4784 ± 895 | Hurt |
| replace autoregressive AIQN actor with non-autoregressive IQN (ablation) | GAC · HalfCheetah-v2 | maximal attained value: 9300 ± 515 -> 6807 ± 98 | Hurt |
| increase the number of samples used to approximate the target policy (ablation) | GAC · MuJoCo control tasks | stability and overall performance | Helped |
| combine uniform action sampling with sampling from the delayed policy (ablation) | GAC target-policy estimation · MuJoCo control tasks | overall performance | Helped |
| use a Uniform target distribution over improving actions (ablation) | GAC · MuJoCo control tasks | performance | Hurt |
Closest papers by meaning
- CTD4 – a Deep Continuous Distributional Actor-Critic Agent with a Kalman Fusion of Multiple Critics (AAAI 2025 · similarity 0.89)
- Wasserstein Policy Optimization (ICML 2025 · similarity 0.88)
- Actor-Free Continuous Control via Structurally Maximizable Q-Functions (NeurIPS 2025 · similarity 0.88)
- A Theoretical Justification for Asymmetric Actor-Critic Algorithms (ICML 2025 · similarity 0.88)
- On the Convergence of Continuous Single-timescale Actor-critic (ICML 2025 · similarity 0.88)
- Langevin Soft Actor-Critic: Efficient Exploration through Uncertainty-Driven Critic Learning (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.