Research map / Reinforcement learning / Policy gradient and actor-critic
CaRe-BN: Precise Moving Statistics for Stabilizing Spiking Neural Networks in Reinforcement Learning
Zijie Xu, Xinyu Shi, Yiting Dong, Zihan Huang, Zhaofei Yu
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 |
|---|---|---|---|
| add CaRe-BN to SNN (ablation) | CLIF SNN · MuJoCo (DDPG) | APG vs vanilla SNN: 0.00% -> 22.62% | Improved |
| add CaRe-BN to SNN | CLIF SNN · MuJoCo (TD3) | APG vs vanilla SNN: 0.00% -> 15.20% | Improved |
| add CaRe-BN to SNN | DN SNN · MuJoCo (TD3) | APG vs vanilla SNN: 0.00% -> 4.37% | Improved |
| add CaRe-BN to SNN | LIF SNN · MuJoCo (DDPG) | APG vs vanilla SNN: 0.00% -> 8.24% | Improved |
| use CaRe-BN instead of other SNN BN variants (TABN) | CLIF SNN, TD3 · MuJoCo 5 envs | APG vs ANN: 3.25% -> 5.90% | Improved |
| use CaRe-BN instead of ILC-SAN architecture | CLIF SNN, TD3 · MuJoCo 5 envs | APG vs ANN: 4.64% -> 5.90% | Improved |
| use CaRe-BN on LIF SNN versus best other BN (tdBN) | LIF SNN, TD3 · MuJoCo 5 envs | APG vs ANN: 1.65% -> 3.92% | Improved |
| add CaRe-BN to ANN agents | ANN, TD3 · MuJoCo 5 envs | APG: 0.00% -> -0.69% (TD3); 0.00% -> 0.33% (DDPG) | No clear effect |
| combine confidence-adaptive update (Ca-BN) and re-calibration (Re-BN) (ablation) | CaRe-BN · MuJoCo (CLIF, TD3) | normalized max performance | Improved |
| increase SNN simulation time steps from 2 to 8 (ablation) | CaRe-BN CLIF TD3 · MuJoCo 5 envs | APG vs ANN: -23.80% -> 6.32% | Improved |
| add CaRe-BN to DDPG CLIF SNN on HalfCheetah | CLIF SNN, DDPG · HalfCheetah | max return: 8205 -> 7972 | Hurt |
| use CaRe-BN instead of ILC-SAN on HalfCheetah | CLIF SNN, TD3 · HalfCheetah | max return: 9222 -> 9563; Ant 5584 -> 5373 | Mixed |
Closest papers by meaning
- Adaptive Surrogate Gradients for Sequential Reinforcement Learning in Spiking Neural Networks (NeurIPS 2025 · similarity 0.89)
- Proxy Target: Bridging the Gap Between Discrete Spiking Neural Networks and Continuous Control (NeurIPS 2025 · similarity 0.88)
- Decision SpikeFormer: Spike-Driven Transformer for Decision Making (CVPR 2025 · similarity 0.87)
- Signal-Adaptive Trust Regions for Gradient-Free Optimization of Recurrent Spiking Neural Networks (NeurIPS 2026 · similarity 0.86)
- Sign Gradient Descent-based Neuronal Dynamics: ANN-to-SNN Conversion Beyond ReLU Network (ICML 2024 · similarity 0.86)
- NDOT: Neuronal Dynamics-based Online Training for Spiking Neural Networks (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.
