Research map / Probabilistic and scientific modeling / Dynamical systems
Energy-based Autoregressive Generation for Neural Population Dynamics
Ningling Ge, Sicheng Dai, Yu Zhu, Shan Yu
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 strictly proper energy score (alpha<2) instead of alpha=2 (ablation) | EAG energy loss · MC Maze | DKL population spike count histogram: 0.0541 -> 0.0014 | Improved |
| use strictly proper energy score (alpha<2) instead of alpha=2 (ablation) | EAG energy loss · MC Maze | RMSE mean ISI: 0.051 -> 0.024 | Improved |
| vary alpha within [1,2) (ablation) | EAG energy loss · MC Maze | RMSE mean ISI: 0.022 to 0.033 | No clear effect |
| use energy loss instead of MSE in autoregressive transformer (ablation) | EAG · MC Maze | generation quality | Improved |
| replace diffusion sampling (LDNS-1000) with EAG-32 single-pass generation | LDNS · MC Maze | latency: 330.64s -> 10.29s | Improved |
| replace LDNS-1000 with EAG-32 | LDNS · MC Maze | RMSE mean ISI: 32.4% gain | Improved |
| replace LDNS with EAG | LDNS · MC Maze | RMSE mean ISI: 0.037 -> 0.024 | Improved |
| replace LDNS with EAG | LDNS · Area2 Bump | RMSE pairwise corr: 0.0076 -> 0.0075 | No clear effect |
| add EAG augmentation for decoder training | NDT · MC Maze | behavior decoding: 0.776 -> 0.838 | Improved |
| add EAG augmentation for decoder training | AutoLFADS · MC Maze | behavior decoding: 0.906 -> 0.911 | Improved |
| scale EAG augmentation from 2x to 4x (ablation) | NDT · Area2 Bump | behavior decoding: 0.621 -> 0.631 | No clear effect |
| condition on unseen velocity trajectories using EAG vs LDNS | LDNS · MC Maze velocity decoding | R2: 0.65 -> 0.89 | Improved |
Closest papers by meaning
- Latent Diffusion for Neural Spiking Data (NeurIPS 2024 · similarity 0.87)
- BLEND: Behavior-guided Neural Population Dynamics Modeling via Privileged Knowledge Distillation (ICLR 2025 · similarity 0.87)
- A Scalable, Causal, and Energy Efficient Framework for Neural Decoding with Spiking Neural Networks (NeurIPS 2025 · similarity 0.86)
- NDOT: Neuronal Dynamics-based Online Training for Spiking Neural Networks (ICML 2024 · similarity 0.86)
- A scalable generative model for dynamical system reconstruction from neuroimaging data (NeurIPS 2024 · similarity 0.86)
- Neuroformer: Multimodal and Multitask Generative Pretraining for Brain Data (ICLR 2024 · similarity 0.86)
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.
