Research map / Probabilistic and scientific modeling / Causal discovery
Training a Resilient Q-network against Observational Interference
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 |
|---|---|---|---|
| Train CIQ with interference labels and a latent causal state, then infer interference at test time | DQN · Vector Cartpole under Gaussian and adversarial interference | AC-Rate: 82.10% -> 99.61%; 72.15% -> 98.52%; 69.74% -> 98.12%; 62.23% -> 99.52%; 9.68% -> 98.52%; 1.22% -> 98.10% | Helped |
| Train CIQ with interference labels and a latent causal state, then infer interference at test time | DQN · Vector Cartpole under Gaussian and adversarial interference | CLEVER-Q: 0.176 -> 0.221; 0.130 -> 0.235; 0.109 -> 0.232; 0.169 -> 0.248; 0.171 -> 0.236; 0.052 -> 0.230 | Helped |
| Train CIQ at a different interference level from the test level | CIQ · Cartpole, noise train/test 0.3/0.1 through 0.3/0.5 | Performance: 195.0; 195.0; 195.0; 195.0; 185.7 | Mixed |
| Train CIQ on Gaussian and adversarial interference together | CIQ · Env1, Gaussian, adversarial, and mixed Gaussian plus adversarial test interference | Performance: 195.0; 195.0; 195.0 | Helped |
| Train on one interference type and test on the other type | CIQ · Env1, Gaussian-to-adversarial and adversarial-to-Gaussian | Performance: 195.1 -> 154.2; 195.0 -> 153.9 | Hurt |
Closest papers by meaning
- Confounding Robust Deep Reinforcement Learning: A Causal Approach (NeurIPS 2025 · similarity 0.89)
- Deep Double Q-learning (NeurIPS 2026 · similarity 0.87)
- Simplifying Deep Temporal Difference Learning (ICLR 2025 · similarity 0.87)
- Adaptive $Q$-Network: On-the-fly Target Selection for Deep Reinforcement Learning (ICLR 2025 · similarity 0.86)
- Turning Sand to Gold: Recycling Data to Bridge On-Policy and Off-Policy Learning via Causal Bound (NeurIPS 2025 · similarity 0.85)
- Universal Approximation Theorem of Deep Q-Networks (ICML 2025 · similarity 0.84)
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.