Research map / Probabilistic and scientific modeling / Dynamical systems
Scalable Bayesian GPFA with automatic relevance determination and discrete noise models
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 |
|---|---|---|---|
| add shift of M1 spike times by +100 ms before fitting bGPFA (ablation) | bGPFA on M1+S1 · Primate reaching | variance explained (%): 68.56% ± 0.09 to 69.81% ± 0.06 | Helped |
| add shift of M1 spike times by +100 ms (ablation) | bGPFA on M1+S1 · Primate reaching | ELBO per neuron: -34,637.0 ± 0.7 to -34,631.1 ± 0.6 | Helped |
| fit M1 and S1 jointly instead of separately | bGPFA · Primate reaching | kinematic prediction | Helped |
| decode from bGPFA latents instead of Gaussian-smoothed spikes | linear kinematic decoder · Primate reaching | kinematic decoding | Helped |
| use bGPFA latents instead of factor analysis for reaction-time prediction | distance to preparatory state · Primate reaching | Pearson rho: 0.21 -> 0.45 | Helped |
| use bGPFA latents instead of raw data for reaction-time prediction | distance to preparatory state · Primate reaching | Pearson rho: 0.002 -> 0.45 | Helped |
| add ARD to bGPFA with D=10 (ablation) | bGPFA · Synthetic data | ELBO / MSE | Helped |
| increase latent dimensionality beyond 3 | FA and GPFA · Synthetic data | cross-validated MSE | Hurt |
| use circulant parameterization instead of diagonal Lambda (ablation) | bGPFA · Primate reaching | qualitative results | No clear effect |
| use bGPFA instead of factor analysis or GPFA for preparatory similarity | latent similarity analysis · Primate reaching | structure of similarity | Helped |
Closest papers by meaning
- Conditionally-Conjugate Gaussian Process Factor Analysis for Spike Count Data via Data Augmentation (ICML 2024 · similarity 0.91)
- Multi-modal Gaussian Process Variational Autoencoders for Neural and Behavioral Data (ICLR 2024 · similarity 0.90)
- Modeling Latent Neural Dynamics with Gaussian Process Switching Linear Dynamical Systems (NeurIPS 2024 · similarity 0.89)
- Modeling Dynamic Neural Activity by combining Naturalistic Video Stimuli and Stimulus-independent Latent Factors (NeurIPS 2025 · similarity 0.87)
- Modeling Neural Activity with Conditionally Linear Dynamical Systems (NeurIPS 2025 · similarity 0.87)
- Probabilistic Decomposed Linear Dynamical Systems for Robust Discovery of Latent Neural Dynamics (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.