Research map / Architectures and efficiency / Spiking networks
Strategy and Benchmark for Converting Deep Q-Networks to Event-Driven Spiking Neural Networks
Weihao Tan, Devdhar Patel, Robert Kozma
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 end-of-simulation membrane potential to firing rate (robust firing rate) (ablation) | Converted DQN-to-SNN agent · Beam Rider, percentile 99.99 | conversion rate: 77.7 ± 0.9% -> 90.6 ± 0.5% | Improved |
| add end-of-simulation membrane potential to firing rate (robust firing rate) (ablation) | Converted DQN-to-SNN agent · Crazy Climber, percentile 99.99 | conversion rate: 45.3 ± 3.1% -> 82.5 ± 3.1% | Improved |
| increase parameter-normalization percentile from 99.9 to 99.99 (ablation) | Converted DQN-to-SNN agent · Star Gunner | score: 54692.0 ± 9963.3 -> 30580.0 ± 15472.6 | Hurt |
| increase simulation time (ablation) | Converted DQN-to-SNN agent · Breakout, James Bond, Star Gunner, Video Pinball | conversion rate | Improved |
| increase parameter-normalization percentile (ablation) | Converted DQN-to-SNN agent · Breakout and James Bond | conversion rate | Improved |
| tune parameter-normalization percentile, with a best point rather than always increasing it (ablation) | Converted DQN-to-SNN agent · Star Gunner and Video Pinball | score | Mixed |
| convert DQN to SNN using the proposed method | DQN · 17 Atari games | score | Mixed |
Closest papers by meaning
- Optimal ANN-SNN Conversion for High-accuracy and Ultra-low-latency Spiking Neural Networks (ICLR 2022 · similarity 0.89)
- QP-SNN: Quantized and Pruned Spiking Neural Networks (ICLR 2025 · similarity 0.88)
- Inference-Scale Complexity in ANN-SNN Conversion for High-Performance and Low-Power Applications (CVPR 2025 · similarity 0.88)
- Faster and Stronger: When ANN-SNN Conversion Meets Parallel Spiking Calculation (ICML 2025 · similarity 0.86)
- Adaptive Calibration: A Unified Conversion Framework of Spiking Neural Networks (AAAI 2025 · similarity 0.86)
- Are Conventional SNNs Really Efficient? A Perspective from Network Quantization (CVPR 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.
