Research map / Probabilistic and scientific modeling / Dynamical systems
Modeling Electrical Motor Dynamics Using Encoder-Decoder with Recurrent Skip Connection
Sagar Verma, Nicolas Henwood, Marc Castella, Francois Malrait, Jean-Christophe Pesquet
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 encoder-decoder skip connections without recurrence (ablation) | Deep encoder-decoder · Simulated validation set | MAE: 81.21 -> 28.96 | Improved |
| Add unidirectional recurrent skip connections (ablation) | Encoder-decoder with skip connections · Simulated validation set | MAE: 28.96 -> 28.18 | Improved |
| Make recurrent skip connections bidirectional (ablation) | RNN-Skip · Simulated validation set | MAE: 28.18 -> 27.96 | Improved |
| Make recurrent skip connections bidirectional (ablation) | RNN-Skip · Simulated validation set | R²: 0.43 -> 0.41 | Hurt |
| Diagonalize bidirectional recurrent skip weights (ablation) | BiRNN-Skip · Simulated validation set | SMAPE: 3.31% -> 1.09% | Improved |
| Diagonalize bidirectional recurrent skip weights (ablation) | BiRNN-Skip · Simulated validation set | R²: 0.41 -> 0.95 | Improved |
| Replace MSE with TV-weighted MSE (ablation) | DiagBiRNN-Skip · Simulated validation set, i_d | SMAPE: 0.9702% -> 0.4619% | Improved |
| Replace MSE with TV-weighted MSE (ablation) | DiagBiRNN-Skip · Simulated validation set, i_q | SMAPE: 2.3925% -> 1.9014% | Improved |
| Replace MSE with TV-weighted MSE (ablation) | DiagBiRNN-Skip · Simulated validation set, τ_em | SMAPE: 1.5839% -> 0.9193% | Improved |
| Fine-tune the simulated-data model on raw sensor data | DiagBiRNN-Skip · Raw sensor test set, i_d | MAE: 39.8392 -> 35.3167 | Improved |
| Fine-tune the simulated-data model on raw sensor data | DiagBiRNN-Skip · Raw sensor test set, i_q | MAE: 47.3818 -> 42.9472 | Improved |
Closest papers by meaning
- Parallel-in-Time Training of Recurrent Neural Networks for Dynamical Systems Reconstruction (NeurIPS 2026 · similarity 0.83)
- Learning Options for Compositional Motor Control with Adapter Banks (NeurIPS 2026 · similarity 0.83)
- Quasi-Linear ICA for Motor Unit Decomposition during Dynamic Contractions (NeurIPS 2026 · similarity 0.82)
- Almost-Linear RNNs Yield Highly Interpretable Symbolic Codes in Dynamical Systems Reconstruction (NeurIPS 2024 · similarity 0.82)
- BRAID: Input-driven Nonlinear Dynamical Modeling of Neural-Behavioral Data (ICLR 2025 · similarity 0.82)
- Recurrent neural networks: vanishing and exploding gradients are not the end of the story (NeurIPS 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.
