Research map / Probabilistic and scientific modeling / Dynamical systems
Training stochastic stabilized supralinear networks by dynamics-neutral growth
Wayne WM Soo, Máté Lengyel
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 |
|---|---|---|---|
| shuffle weights within E/I quadrants of trained SSN (ablation) | trained SSN 50:50 · MNIST | proportion unstable: 0.0 -> 0.790 | Hurt |
| shuffle weights within E/I quadrants of trained SSN (ablation) | trained SSN 80:20 · MNIST | proportion unstable: 0.0 -> 0.794 | Hurt |
| replace trained weights with random Gaussian weights (ablation) | trained SSN · MNIST | proportion unstable: 0.0 -> 1.0 | Hurt |
| shuffle weights of trained general SSN (ablation) | general SSN · GSM inference | proportion unstable: 0.0 -> 0.446 | Hurt |
| use SSN instead of MLP readout network | SSN 50:50 · MNIST | test accuracy: 0.974 -> 0.949 | Hurt |
| use SSN instead of logistic regression | SSN 50:50 · MNIST | test accuracy: 0.914 -> 0.949 | Improved |
| use SSN instead of logistic regression on encoded-80 | SSN 80:20 · MNIST | test accuracy: 0.922 -> 0.952 | Improved |
| replace Gaussian-profile ring SSN with unconstrained SSN | SSN 50:50 · MNIST | test accuracy: 0.223 -> 0.949 | Improved |
| train ring SSN with our method instead of Gaussian ring SSN [29] | ring SSN · ring GSM | normalized cost: 1.0 -> 0.158 | Improved |
| use general unconstrained SSN instead of Gaussian ring SSN | SSN · CIFAR-10-optimized GSM | normalized cost: ∼104 -> 0.745 | Improved |
| change E:I ratio from 50:50 to 80:20 | SSN · MNIST | test accuracy: 0.949 -> 0.952 | No clear effect |
Closest papers by meaning
- CORNN: Convex optimization of recurrent neural networks for rapid inference of neural dynamics (NeurIPS 2023 · similarity 0.89)
- Unconditional stability of a recurrent neural circuit implementing divisive normalization (NeurIPS 2024 · similarity 0.87)
- Recurrent neural network dynamical systems for biological vision (NeurIPS 2024 · similarity 0.86)
- Learning the Plasticity: Plasticity-Driven Learning Framework in Spiking Neural Networks (NeurIPS 2025 · similarity 0.86)
- Scaling Laws and Tradeoffs in Recurrent Networks of Expressive Neurons (NeurIPS 2026 · similarity 0.86)
- Temporal Flexibility in Spiking Neural Networks: Towards Generalization Across Time Steps and Deployment Friendliness (ICLR 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.
