Research map / Probabilistic and scientific modeling / Dynamical systems
A new inference approach for training shallow and deep generalized linear models of noisy interacting neurons
Gabriel Mahuas, Giulio Isacchini, Olivier Marre, Ulisse Ferrari, Thierry Mora
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 |
|---|---|---|---|
| two-step inference instead of whole log-likelihood maximization | retinal GLM · Moving bars | noise correlation CoD: 0.55 -> 0.91 | Improved |
| two-step inference instead of whole log-likelihood maximization | retinal GLM · Moving bars | PSTH Pearson correlation: 0.71 -> 0.72 | No clear effect |
| two-step inference instead of whole log-likelihood maximization | retinal GLM · Moving bars | self-excitation frequency: 10% -> 0% | Improved |
| two-step inference instead of whole log-likelihood maximization | retinal GLM · Checkerboard | noise correlation CoD: 0.94 -> 0.95 | No clear effect |
| use couplings inferred on moving bars with two-step instead of whole log-l | retinal GLM stim filter from checkerboard · Checkerboard | self-excitation frequency: 93% -> 0% | Improved |
| transfer couplings across stimuli with two-step inference | two-step GLM · Checkerboard | noise correlation CoD: 0.95 -> 0.84 | Hurt |
| transfer couplings across stimuli with two-step inference | two-step GLM · Checkerboard | PSTH Pearson correlation: 0.81 -> 0.81 | No clear effect |
| replace linear stimulus filter with deep time-distributed CNN in two-step model | two-step GLM · Checkerboard | PSTH Pearson correlation: 0.81 -> 0.87 | Improved |
| replace linear stimulus filter with deep CNN | two-step GLM · Checkerboard | noise correlation CoD: 0.95 -> 0.93 | No clear effect |
| estimate stimulus filters in presence of fixed couplings instead of the mean-correction approximation (ablation) | two-step GLM · retina data | model accuracy | No clear effect |
Closest papers by meaning
- Natural image synthesis for the retina with variational information bottleneck representation (NeurIPS 2022 · similarity 0.88)
- From deep learning to mechanistic understanding in neuroscience: the structure of retinal prediction (NeurIPS 2019 · similarity 0.87)
- Identifying interactions across brain areas while accounting for individual-neuron dynamics with a Transformer-based variational autoencoder (NeurIPS 2025 · similarity 0.86)
- Scalable inference of functional neural connectivity at submillisecond timescales (NeurIPS 2025 · similarity 0.86)
- Benchmarking Out-of-Distribution Generalization Capabilities of DNN-based Encoding Models for the Ventral Visual Cortex. (NeurIPS 2024 · similarity 0.85)
- A Differentiable Partially Observable Generalized Linear Model with Forward-Backward Message Passing (ICML 2024 · 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.
