Research map / Probabilistic and scientific modeling / Dynamical systems
Scaling the Poisson GLM to massive neural datasets through polynomial approximations
David Zoltowski, Jonathan W Pillow
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 paGLM-2 quadratic approximation instead of exact MAP | Poisson GLM · Neuropixels 10 min, 50 neurons | fitting time: about 3 minutes -> about 3 seconds | Improved |
| Use paGLM-2 instead of exact MAP | Poisson GLM · Neuropixels 10 min | train/held-out log-likelihood | No clear effect |
| Adaptive approximation interval instead of fixed interval (ablation) | paGLM-2 · Neuropixels 10 min | held-out log-likelihood | Improved |
| Select approximation interval adaptively (ablation) | paGLM-2 exponential nonlinearity · RGC data | parameter accuracy | Improved |
| Use quadratic approximate evidence instead of Laplace evidence for ridge selection | paGLM-2 · RGC 100 weights | time to select ridge: 30 -> 0.7 seconds | Improved |
| Use paGLM-4 (fourth order) instead of paGLM-2 | paGLM · RGC exponential nonlinearity | filter estimate accuracy | No clear effect |
| Use paGLM-2 instead of exact ML at small data (ablation) | Poisson GLM · Simulated data | MSE | Improved |
| Use softplus instead of exponential nonlinearity | paGLM-2 · RGC data | robustness to approximation interval | Improved |
| Store approximate sufficient statistics instead of design matrix | paGLM-2 · Neuropixels 41 min | storage: reduced by a factor of ≈40 | Improved |
Closest papers by meaning
- Scalable inference of functional neural connectivity at submillisecond timescales (NeurIPS 2025 · similarity 0.93)
- Identifying interactions across brain areas while accounting for individual-neuron dynamics with a Transformer-based variational autoencoder (NeurIPS 2025 · similarity 0.88)
- Gaussian process based nonlinear latent structure discovery in multivariate spike train data (NeurIPS 2017 · similarity 0.88)
- A Differentiable Partially Observable Generalized Linear Model with Forward-Backward Message Passing (ICML 2024 · similarity 0.87)
- Spike Distance Function as a Learning Objective for Spike Prediction (ICML 2024 · similarity 0.86)
- Modeling Latent Neural Dynamics with Gaussian Process Switching Linear Dynamical Systems (NeurIPS 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.
