Research map / Reinforcement learning / Policy gradient and actor-critic
DNA: Proximal Policy Optimization with a Dual Network Architecture
Matthew Aitchison, Penny Sweetser
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| remove distillation (ablation) | DNA · Atari-5 | score: 252 -> 226 | Hurt |
| set policy, value, and distillation mini-batch sizes to 512 (ablation) | DNA · Atari-5 | score: 252 -> 195 | Hurt |
| use equal λ values of 0.95 for policy and value (ablation) | DNA · Atari-5 | score: 252 -> 191 | Hurt |
| replace DNA's dual-network design and novel components with a basic single-network PPO setup (ablation) | DNA · Atari-5 | score: 252 -> 81 | Hurt |
| use a lower λπ of 0.8 while retaining λV=0.95 | DNA · Atari-3 validation | score | Improved |
| use fewer than three policy epochs, selecting Eπ=2 (ablation) | DNA · Atari-3 validation | score | Improved |
| increase value training to four epochs (ablation) | DNA · Atari-3 validation | score | Hurt |
| remove the distillation phase (ablation) | DNA · Atari-3 validation | score | Hurt |
| use random-projection distillation targets (ablation) | DNA · Atari-3 validation | score | Improved |
| use value-estimate distillation targets (ablation) | DNA · Atari-3 validation | score | Improved |
| use feature-matching distillation targets (ablation) | DNA · Atari-3 validation | score | Hurt |
| increase PPO policy epochs from one to two (ablation) | PPO · Atari-5 | score | Hurt |
Closest papers by meaning
- Learning Value Functions in Deep Policy Gradients using Residual Variance (ICLR 2021 · similarity 0.87)
- XQC: Well-conditioned Optimization Accelerates Deep Reinforcement Learning (ICLR 2026 · similarity 0.87)
- Wasserstein Policy Optimization (ICML 2025 · similarity 0.86)
- Value Improved Actor Critic Algorithms (NeurIPS 2025 · similarity 0.86)
- ReLU to the Rescue: Improve Your On-Policy Actor-Critic with Positive Advantages (ICML 2024 · similarity 0.86)
- In value-based deep reinforcement learning, a pruned network is a good network (ICML 2024 · similarity 0.86)
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.
