Research map / Architectures and efficiency / RNNs and normalization
Functional building blocks of neural networks: from network motifs to collective dynamics
Jian Zhang, Yue Sun, Wangzi Yao, Tielin Zhang
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 Level-1 motif regularization to a recurrent network | RNN · PDM | mean accuracy: 87.1 -> 91.9 | Improved |
| add Level-1 motif regularization to a recurrent network | RNN · TIDIGITS | mean accuracy: 40.9 -> 45.0 | Improved |
| add Level-1 motif regularization to a recurrent network | RNN · Sequential MNIST | mean accuracy: 61.7 -> 67.0 | Improved |
| use Level-1 motifs instead of Level-2 motifs (ablation) | motif RNN · MNIST noise robustness | mean accuracy: 9.5±1.2% | Improved |
| use Level-1 motifs instead of Level-3 motifs (ablation) | motif RNN · PDM | mean accuracy: 4.7±0.6% | Improved |
| use Level-1 motifs instead of Level-3 motifs (ablation) | motif RNN · TIDIGITS | mean accuracy: 5.3±0.7% | Improved |
| use Level-3 motifs instead of Level-1 motifs (ablation) | motif RNN · Ant-v2 (2M steps) | score: +384.5 ± 125.1 | Improved |
| use Level-3 motifs instead of Level-1 motifs (ablation) | motif RNN · Ant-v2 (5M steps) | score: +417.5 ± 164.3 | Improved |
| use Level-3 motifs instead of Level-1 motifs (ablation) | motif RNN · InvertedPendulum-v2 | convergence speed: 31.8 ± 9.2% faster | Improved |
| add Level-3 motif regularization to a recurrent network | RNN · Ant-v2 (5M steps) | score: 2882.1 -> 4258.0 | Improved |
| use motif regularization instead of LSTM (ablation) | RNN · Sequential MNIST | mean accuracy: 68.19 -> 67.0 | No clear effect |
| use very small relaxation beta (ablation) | motif regularization · PDM | accuracy | Hurt |
Closest papers by meaning
- Learning dynamic representations of the functional connectome in neurobiological networks (ICLR 2024 · similarity 0.86)
- Mechanistic Interpretability of RNNs emulating Hidden Markov Models (NeurIPS 2025 · similarity 0.86)
- Partial observation can induce mechanistic mismatches in data-constrained models of neural dynamics (NeurIPS 2024 · similarity 0.86)
- How connectivity structure shapes rich and lazy learning in neural circuits (ICLR 2024 · similarity 0.86)
- Discovering plasticity rules that organize and maintain neural circuits (NeurIPS 2024 · similarity 0.86)
- Self-Supervised Discovery of Neural Circuits in Spatially Patterned Neural Responses with Graph Neural Networks (NeurIPS 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.
