Research map / Probabilistic and scientific modeling / Dynamical systems
Enabling hyperparameter optimization in sequential autoencoders for spiking neural data
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 |
|---|---|---|---|
| Use sample validation instead of holding out entire trials | LFADS on the synthetic RNN dataset · Synthetic RNN dataset | correspondence between validation loss and firing-rate R² | Helped |
| Add coordinated dropout during training | LFADS on the synthetic RNN dataset · Synthetic RNN dataset | correspondence between standard validation loss and firing-rate R² | Helped |
| Use coordinated dropout rather than sample validation alone | LFADS on the synthetic RNN dataset · Synthetic RNN dataset | reliability of validation loss as a performance measure | Helped |
| Apply coordinated dropout during training | Linear autoencoder on noise-corrupted low-dimensional data · Synthetic low-dimensional data | reconstruction loss on noise-free data | Helped |
| Use population-based hyperparameter optimization with coordinated dropout instead of fixed hyperparameters | LFADS · Monkey J Maze, reduced training datasets | hand-velocity decoding R² | Helped |
| Use population-based hyperparameter optimization with coordinated dropout instead of fixed hyperparameters | LFADS · Monkey J Maze, full training dataset | hand-velocity decoding R² | No clear effect |
| Use the fixed hyperparameters selected for Monkey J Maze instead of the CursorJump hyperparameters | LFADS · Monkey J Maze, reduced training datasets | hand-velocity decoding R² | Hurt |
| Increase coordinated-dropout keep ratio so more samples are seen as input | LFADS · Datasets with few observed dimensions relative to latent dimensionality | latent-structure inference | Mixed |
Closest papers by meaning
- A Scalable, Causal, and Energy Efficient Framework for Neural Decoding with Spiking Neural Networks (NeurIPS 2025 · similarity 0.88)
- Latent Diffusion for Neural Spiking Data (NeurIPS 2024 · similarity 0.88)
- Inference of Neural Dynamics Using Switching Recurrent Neural Networks (NeurIPS 2024 · similarity 0.87)
- P-SPIKESSM: HARNESSING PROBABILISTIC SPIKING STATE SPACE MODELS FOR LONG-RANGE DEPENDENCY TASKS (ICLR 2025 · similarity 0.87)
- NDOT: Neuronal Dynamics-based Online Training for Spiking Neural Networks (ICML 2024 · similarity 0.87)
- LongSpike: Fractional Order Spiking State Space Models for Efficient Long Sequence Learning (NeurIPS 2026 · 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.