Research map / Architectures and efficiency / RNNs and normalization
Biologically Plausible Learning via Bidirectional Spike-Based Distillation
Yifei Wang, Zhangyanxun, Changze Lv, Yiyang Lu, Jingwen Xu, Xiaohua Wang, Di Yu, Xin Du, Xuanjing Huang, Xiaoqing Zheng
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 |
|---|---|---|---|
| Replace ReCo loss with MSE for layer-wise voltage alignment (ablation) | BSD · CIFAR-100 (CNN) | accuracy: 53.48% -> 1.58% | Hurt |
| Replace ReCo loss with InfoNCE (ablation) | BSD · CIFAR-100 (CNN) | accuracy: 53.48% -> 38.06% | Hurt |
| Replace ReCo loss with InfoNCE (ablation) | BSD · CIFAR-10 (CNN) | accuracy: 84.13% -> 72.38% | Hurt |
| Replace ReCo loss with InfoNCE (ablation) | BSD · SVHN (CNN) | accuracy: 90.81% -> 83.27% | Hurt |
| Set ReCo penalty weight lambda to 0 (ablation) | BSD CNN · SVHN | accuracy | Hurt |
| Set ReCo penalty lambda to 0.6 (ablation) | BSD CNN · SVHN | accuracy | Improved |
| Set firing threshold to 0.6 rather than lower or higher (ablation) | BSD · SVHN | accuracy | Improved |
| Use spiking neurons instead of continuous activations under backpropagation | BP on SNN vs ANN · CIFAR-10 (MLP) | accuracy: 57.65% -> 49.95% | Hurt |
| Use BSD instead of Dendritic Localized Learning | CNN · CIFAR-100 | accuracy: 38.60% -> 53.48% | Improved |
| Use BSD instead of backprop on ANN for time-series | RNN · Electricity | MSE: 0.175 -> 0.165 | Improved |
| Use BSD instead of backprop on ANN | autoencoder · CIFAR-10 generation | FID: 127.34 -> 168.12 | Hurt |
| Use BSD instead of fully spiking VAE | autoencoder · MNIST generation | FID: 97.06 -> 72.39 | Improved |
Closest papers by meaning
- Temporal Separation with Entropy Regularization for Knowledge Distillation in Spiking Neural Networks (CVPR 2025 · similarity 0.87)
- Synergy Between the Strong and the Weak: Spiking Neural Networks are Inherently Self-Distillers (NeurIPS 2025 · similarity 0.87)
- Cross-Modal Representational Knowledge Distillation for Enhanced Spike-informed LFP Modeling (NeurIPS 2025 · similarity 0.86)
- Learning the Plasticity: Plasticity-Driven Learning Framework in Spiking Neural Networks (NeurIPS 2025 · similarity 0.86)
- Forward Learning with Top-Down Feedback: Empirical and Analytical Characterization (ICLR 2024 · similarity 0.86)
- Efficient ANN-Guided Distillation: Aligning Rate-based Features of Spiking Neural Networks through Hybrid Block-wise Replacement (CVPR 2025 · 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.
