Research map / Time series / Causal and generative time series
Weight-Space Linear Recurrent Neural Networks
Roussel Desmond Nzoyem, Nawid Keshtmand, Enrique Crespo-Fernandez, Idriss TSAYEM, Raul Santos-Rodriguez, David A.W. Barton, Tom Deakin
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 physical prior (mass-spring-damper form) in root network | WARP · MSD | MSE: 0.94 -> 0.03 | Improved |
| add physical prior in root network | WARP · MSD-Zero | MSE: 0.32 -> 0.04 | Improved |
| embed sine formula with predicted phase in root network | WARP · SINE* (Small split) | MSE: 2.77 -> 0.62 | Improved |
| use weight-space recurrence instead of GRU | GRU baseline · LV | MSE: 5.83 -> 4.72 | Improved |
| use weight-space recurrence instead of GRU | GRU baseline · CelebA (L=300) | BPD: 60.39 -> -0.043 | Improved |
| use weight-space recurrence instead of Transformer | Transformer (TST) · SINE* (Small split) | MSE: 1728 -> 2.77 | Improved |
| use WARP instead of graph-based STIDGCN | STIDGCN · PEMS08 | MAE: 13.45 -> 6.59 | Improved |
| use WARP instead of S4 | S4 · MNIST (L=600) | BPD: 0.406 -> 0.416 | Hurt |
| use WARP instead of Transformer | Transformer (TST) · MSD | MSE: 0.34 -> 0.94 | Hurt |
| use WARP instead of LinOSS | LinOSS · UEA EigenWorms | accuracy: 95.0 -> 70.93 | Hurt |
| use WARP instead of LRU | LRU · UEA Heartbeat | accuracy: 78.1 -> 80.65 | Improved |
Closest papers by meaning
- RNNs perform task computations by dynamically warping neural representations (NeurIPS 2025 · similarity 0.87)
- RECAST: Reparameterized, Compact weight Adaptation for Sequential Tasks (ICLR 2025 · similarity 0.87)
- Learning dynamics in linear recurrent neural networks (ICML 2025 · similarity 0.86)
- Recurrent neural networks: vanishing and exploding gradients are not the end of the story (NeurIPS 2024 · similarity 0.86)
- Learning Useful Representations of Recurrent Neural Network Weight Matrices (ICML 2024 · similarity 0.85)
- Understanding the Differences in Foundation Models: Attention, State Space Models, and Recurrent Neural Networks (NeurIPS 2024 · similarity 0.85)
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.
