Research map / Probabilistic and scientific modeling / Dynamical systems
Deep inference of latent dynamics with spatio-temporal super-resolution using selective backpropagation through time
Feng Zhu, Andrew R Sedler, Harrison A Grier, Nauman Ahad, Mark A. Davenport, Matthew Kaufman, Andrea Giovannucci, Chethan Pandarinath
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 |
|---|---|---|---|
| train AutoLFADS with SBTT on masked data instead of fully observed GPFA baseline | AutoLFADS with SBTT · Macaque M1/PMd maze, hand velocity decoding | R^2: 15% | Improved |
| increase fraction of dropped samples up to 80% | AutoLFADS with SBTT · Macaque M1/PMd maze, hand velocity decoding | R^2: 80% | No clear effect |
| replace AutoLFADS with NDT trained with selective backpropagation (ablation) | NDT with selective backprop · Macaque M1/PMd maze, decoding from rates | R^2: 40% | Hurt |
| use sub-frame phase information with SBTT instead of frame-resolution ALFADS or Gaussian smoothing (ablation) | ALFADS-SBTT · Simulated 2P Lorenz, 15Hz | R^2: R^2 ~ 0.8 up to 15Hz | Improved |
| use sub-frame sampling times with SBTT instead of frame-resolution ALFADS (ablation) | ALFADS-SBTT · Mouse 2P calcium imaging, forepaw kinematics | R^2 decoding | Improved |
| use SBTT instead of ALFADS for PSTH match (ablation) | ALFADS-SBTT · Mouse 2P calcium imaging | correlation with PSTH | Improved |
| retrain encoder networks with SBTT on sparse data after full-data training (ablation) | AutoLFADS · Macaque M1/PMd maze, sparse data | hand velocity decoding R^2 | Improved |
Closest papers by meaning
- Dynamical Modeling of Behaviorally Relevant Spatiotemporal Patterns in Neural Imaging Data (ICML 2025 · similarity 0.88)
- NDOT: Neuronal Dynamics-based Online Training for Spiking Neural Networks (ICML 2024 · similarity 0.87)
- Active learning of neural population dynamics using two-photon holographic optogenetics (NeurIPS 2024 · similarity 0.87)
- Modeling Latent Neural Dynamics with Gaussian Process Switching Linear Dynamical Systems (NeurIPS 2024 · similarity 0.87)
- Latent Diffusion for Neural Spiking Data (NeurIPS 2024 · similarity 0.87)
- Dynamical modeling of nonlinear latent factors in multiscale neural activity with real-time 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.
