Research map / Architectures and efficiency / RNNs and normalization
Deep Multi-State Dynamic Recurrent Neural Networks Operating on Wavelet Based Neural Features for Robust Brain Machine Interfaces
Benyamin Allahgholizadeh Haghi, Spencer Kellis, Sahil Shah, Maitreyi Ashok, Luke Bashford, Daniel Kramer, Brian Lee, Charles Liu, Richard Andersen, Azita Emami
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 |
|---|---|---|---|
| add prediction feedback and scheduled sampling to the recurrent decoder (DRNN vs standard recurrent nets) | DRNN · single-day PPC center-out, MWT features | average R2 (Y position) | Improved |
| use mid-frequency wavelet features (MWT) instead of other 11 neural features | DRNN · multi-day, trained on first 20 days of 2015 | R2 | Improved |
| use MWT instead of other features in cross-day analysis (train on one day, test on all others) | DRNN · cross-day, 43 days | average R2 | Improved |
| train on only 10% of single-day data | DRNN · single-day, 2018-04-23 | regression quality (qualitative) | No clear effect |
| use DRNN instead of other linear and nonlinear decoders over many days | DRNN · multi-day, days 21-43 | average R2 | Improved |
| increase number of training days from 1 to 20 | Deep-DRNN and DRNN · multi-day | R2 | Improved |
| use small DRNN (1 layer, 10 nodes, history 10) instead of Deep-DRNN (2 layers, 50 and 25 nodes, history 20) | DRNN · multi-day, varying training days | R2 | No clear effect |
| re-train DRNN with 5-90% of subsequent test days | DRNN · multi-day | R2 | Improved |
| re-train annually versus only once on the first 20 days | DRNN · multi-day | R2 | No clear effect |
| train on first 90% of the same test day versus 20 prior days | DRNN · multi-day | average R2 | Improved |
| use feedback only, feedback with noise instead of neural data, or neural data only at trial start (ablation) | DRNN · center-out predictions | prediction quality (qualitative) | Hurt |
Closest papers by meaning
- Recurrent neural network dynamical systems for biological vision (NeurIPS 2024 · similarity 0.86)
- Exploring the trade-off between deep-learning and explainable models for brain-machine interfaces (NeurIPS 2024 · similarity 0.85)
- Unconditional stability of a recurrent neural circuit implementing divisive normalization (NeurIPS 2024 · similarity 0.84)
- Inference of Neural Dynamics Using Switching Recurrent Neural Networks (NeurIPS 2024 · similarity 0.83)
- Second-order forward-mode optimization of recurrent neural networks for neuroscience (NeurIPS 2024 · similarity 0.83)
- Emergent mechanisms for long timescales depend on training curriculum and affect performance in memory tasks (ICLR 2024 · similarity 0.82)
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.
