Research map / Probabilistic and scientific modeling / Dynamical systems
Reinforcing Neural Network Stability with Attractor Dynamics
Hanming Deng, Yang Hua, Tao Song, Zhengui Xue, Ruhui Ma, Neil Robertson, Haibing Guan
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 RMAN with 256 placement attractors (ablation) | ResNet-44 · CIFAR100 | test error (%): 28.16 -> 27.57 | Improved |
| add RMAN with 256 placement attractors (ablation) | ResNext-29 (8x64d) · CIFAR100 | test error (%): 19.59 -> 18.22 | Improved |
| add RMAN with 256 placement attractors (ablation) | LM-ResNet-110 · CIFAR100 | test error (%): 25.46 -> 24.36 | Improved |
| add RMAN with 256 placement attractors (ablation) | ResNet-20 · CIFAR10 | test error (%): 7.95 -> 7.68 | Improved |
| add RMAN with 16 placement attractors (ablation) | ResNet-44 · CIFAR100 | test error (%): 28.16 -> 27.87 | Improved |
| add RMAN with 64 placement attractors (ablation) | ResNet-44 · CIFAR100 | test error (%): 28.16 -> 28.02 | No clear effect |
| vary number of placement attractors from 16 to 256 (ablation) | RMAN · CIFAR10/CIFAR100 | test error (%) | No clear effect |
| add RMAN to Neural ODE (8 attractors) (ablation) | ODE-Net · CIFAR10 | test error (%): 12.09 -> 11.64 | Improved |
| add RMAN to Neural ODE (16 attractors) (ablation) | ODE-Net · CIFAR100 | test error (%): 41.06 -> 40.54 | Improved |
| add RMAN to Neural ODE (8 attractors) (ablation) | ODE-Net · CIFAR10 | NFE: 26.3 -> 32.4 | Improved |
| add RMAN with 256 placement attractors (ablation) | ResNet-34 · Google Commands | test error (%): 4.02 -> 3.78 | Improved |
| add RMAN with 256 placement attractors (ablation) | ResNet-18 · Google Commands | test error (%): 4.06 -> 3.88 | Improved |
Closest papers by meaning
- Unconditional stability of a recurrent neural circuit implementing divisive normalization (NeurIPS 2024 · similarity 0.86)
- Learning dynamics in linear recurrent neural networks (ICML 2025 · similarity 0.85)
- Feature Learning Dynamics in Infinite-Depth Neural Networks (NeurIPS 2026 · similarity 0.85)
- Three Mechanisms of Feature Learning in a Linear Network (ICLR 2025 · similarity 0.85)
- Sign Gradient Descent-based Neuronal Dynamics: ANN-to-SNN Conversion Beyond ReLU Network (ICML 2024 · similarity 0.85)
- Learning Dynamics of RNNs in Closed-Loop Environments (NeurIPS 2025 · similarity 0.85)
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.
