Research map / Learning theory and algorithms / Prediction and decision losses
Linear Dynamical Systems as a Core Computational Primitive
Shiva Kaul
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 |
|---|---|---|---|
| LDS instead of LSTM for sequential permuted MNIST | LDS · Sequential permuted MNIST | accuracy (%): 92.3% -> 91.8% | No clear effect |
| LDS instead of unfused RNN for step time | LDS · Sequential permuted MNIST | step time (ms): 324ms -> 73ms | Improved |
| LDS with 16,500 parameters instead of LSTM with 68,000 | LDS · Sequential permuted MNIST | parameters: ~68,000 -> ~16,500 | Improved |
| LDS (linear RNN, n=160) instead of LSTM/tanh RNN/uRNN on copy memory | LDS · Copy memory T=2000 | parameters needed to solve: 6500 -> 3380 | Improved |
| LDStack instead of unfused RNNs for runtime | LDStack · Runtime vs sequence length | time (seconds) | Improved |
| LDStack instead of CuDNN LSTM at long sequences | LDStack · Runtime vs sequence length | time (seconds) | Mixed |
| simple tanh RNN on adding problem | Simple RNN · Adding problem | mean-squared error: 0.167 | Hurt |
| LDStack instead of LSTM on adding problem | LDStack · Adding problem | convergence speed | Hurt |
Closest papers by meaning
- Towards Scalable and Stable Parallelization of Nonlinear RNNs (NeurIPS 2024 · similarity 0.88)
- Parallel-in-Time Training of Recurrent Neural Networks for Dynamical Systems Reconstruction (NeurIPS 2026 · similarity 0.86)
- Almost-Linear RNNs Yield Highly Interpretable Symbolic Codes in Dynamical Systems Reconstruction (NeurIPS 2024 · similarity 0.86)
- Predictability Enables Parallelization of Nonlinear State Space Models (NeurIPS 2025 · similarity 0.85)
- SpectraLDS: Provable Distillation for Linear Dynamical Systems (NeurIPS 2025 · similarity 0.85)
- A Memory Efficient Unified Algorithm for Online Learning of Linear Dynamical Systems (NeurIPS 2026 · similarity 0.84)
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.
