Research map / Probabilistic and scientific modeling / Dynamical systems
Dis-inhibitory neuronal circuits can control the sign of synaptic plasticity
Julian Rossbroich, Friedemann Zenke
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 |
|---|---|---|---|
| use linear threshold rule instead of exact inverse rule (1 hidden layer) (ablation) | dis-inhibitory control · MNIST | test accuracy: 97.7 -> 97.1 | Hurt |
| use linear threshold rule instead of exact inverse rule (3 hidden layers) (ablation) | dis-inhibitory control · MNIST | test accuracy: 98.0 -> 96.5 | Hurt |
| use linear threshold rule instead of exact inverse rule (1 hidden layer) (ablation) | dis-inhibitory control · Fashion-MNIST | test accuracy: 89.1 -> 87.6 | Hurt |
| use linear threshold rule instead of exact inverse rule (3 hidden layers) (ablation) | dis-inhibitory control · Fashion-MNIST | test accuracy: 89.1 -> 86.8 | Hurt |
| replace input-dependent Jacobian feedback with average Jacobian feedback (3 hidden layers) (ablation) | dis-inhibitory control, exact inverse · MNIST | test accuracy: 98.0 -> 97.0 | Hurt |
| replace input-dependent Jacobian feedback with average Jacobian feedback (3 hidden layers) (ablation) | dis-inhibitory control, exact inverse · Fashion-MNIST | test accuracy: 89.1 -> 88.2 | Hurt |
| replace input-dependent Jacobian feedback with average Jacobian feedback (1 hidden layer) (ablation) | dis-inhibitory control, exact inverse · MNIST | test accuracy: 97.7 -> 97.8 | No clear effect |
| replace input-dependent Jacobian feedback with average Jacobian feedback (3 hidden layers) (ablation) | dis-inhibitory control, linear threshold · Fashion-MNIST | test accuracy: 86.8 -> 84.6 | Hurt |
| use exact-inverse dis-inhibitory control instead of backprop (1 hidden layer) | MLP · MNIST | test accuracy: 98.1 -> 97.7 | No clear effect |
| use exact-inverse dis-inhibitory control instead of backprop (3 hidden layers) | MLP · Fashion-MNIST | test accuracy: 89.4 -> 89.1 | Hurt |
| use linear threshold rule instead of exact inverse in student-teacher task (ablation) | dis-inhibitory control · student-teacher sine task | MSE loss | No clear effect |
| increase depth from 1 to 3 hidden layers with linear threshold rule (ablation) | dis-inhibitory control · MNIST | test accuracy: 97.1 -> 96.5 | Hurt |
Closest papers by meaning
- Dendritic cortical microcircuits approximate the backpropagation algorithm (NeurIPS 2018 · similarity 0.91)
- Model Based Inference of Synaptic Plasticity Rules (NeurIPS 2024 · similarity 0.89)
- Feedback control guides credit assignment in recurrent neural networks (NeurIPS 2024 · similarity 0.89)
- Discovering plasticity rules that organize and maintain neural circuits (NeurIPS 2024 · similarity 0.87)
- Learning the Plasticity: Plasticity-Driven Learning Framework in Spiking Neural Networks (NeurIPS 2025 · similarity 0.87)
- The Bayesian sampling in a canonical recurrent circuit with a diversity of inhibitory interneurons (NeurIPS 2024 · similarity 0.87)
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.
