Research map / Probabilistic and scientific modeling / Dynamical systems
Latent Equilibrium: A unified learning theory for arbitrarily fast computation with arbitrarily slow neurons
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 prospective neuronal rates instead of instantaneous rates (ablation) | Latent Equilibrium fully connected network · MNIST test | accuracy | Helped |
| Use prospective coding (ablation) | Cortical microcircuit · Synthetic classification | input presentation time for perfect accuracy: 100\,\tau^\text{eff} versus degradation only below 0.1\,\tau^\text{eff} | Helped |
| Use Latent Equilibrium rather than a standard ANN trained with SGD | Fully connected model · MNIST test | classification test error: 1.93 \pm{} 0.14 -> 1.98 \pm{} 0.11 | No clear effect |
| Use Latent Equilibrium rather than a standard ANN trained with SGD | Fully connected model · HIGGS test | classification test error: 27.8 \pm{} 0.4 -> 27.6 \pm{} 0.4 | No clear effect |
| Use Latent Equilibrium rather than a standard ANN trained with backpropagation | Convolutional model · MNIST test | classification test error: 1.08 \pm{} 0.07 -> 1.1 \pm{} 0.1 | No clear effect |
| Use Latent Equilibrium rather than a standard ANN trained with backpropagation | Convolutional model · CIFAR10 test | classification test error: 39.4 \pm{} 5.6 -> 38.0 \pm{} 1.3 | No clear effect |
| Add neuron-local adaptation of membrane time constants | Latent Equilibrium with heterogeneous time constants · Time-constant mismatch | residual mismatch: mismatches below 1\,\textperthousand | Helped |
| Add synaptic filtering | Latent Equilibrium under additive temporal output noise · Noisy network learning | learning reliability | Helped |
Closest papers by meaning
- The Expressive Leaky Memory Neuron: an Efficient and Expressive Phenomenological Neuron Model Can Solve Long-Horizon Tasks. (ICLR 2024 · similarity 0.89)
- Scaling Laws and Tradeoffs in Recurrent Networks of Expressive Neurons (NeurIPS 2026 · similarity 0.89)
- Self-Supervised Discovery of Neural Circuits in Spatially Patterned Neural Responses with Graph Neural Networks (NeurIPS 2025 · similarity 0.88)
- Feedback control guides credit assignment in recurrent neural networks (NeurIPS 2024 · similarity 0.88)
- Partial observation can induce mechanistic mismatches in data-constrained models of neural dynamics (NeurIPS 2024 · similarity 0.88)
- Brain-like Variational Inference (NeurIPS 2025 · 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.