Research map / Time series / Causal and generative time series
Unsupervised Discovery of Temporal Structure in Noisy Data with Dynamical Components Analysis
David Clark, Jesse Livezey, Kristofer Bouchard
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 |
|---|---|---|---|
| Use DCA instead of PCA | Linear dimensionality reduction for time-series data · Noisy embedded Lorenz | Lorenz reconstruction performance (R²) | Improved |
| Use DCA instead of PCA | Motor cortical and hippocampal neural data · M1 and Hippocampus | Behavioral-variable prediction | Improved |
| Use DCA instead of PCA | Temperature forecasting · Temperature | Future-state forecasting | No clear effect |
| Use DCA instead of PCA | Accelerometer forecasting · Accelerometer | Future-state forecasting | Improved |
| Use DCA instead of SFA | Linear dimensionality reduction for real-world time series · M1, Hippocampus, Temperature, and Accelerometer | Decoding or forecasting held-out R² | Improved |
| Increase the past-future window size T (ablation) | DCA · M1, Hippocampus, Temperature, and Accelerometer | Decoding or forecasting held-out R² relative to SFA | Improved |
| Use DCA instead of SFA | Dynamical subspace extraction · Time-irreversible and/or non-Markovian processes | Past-future mutual information | Improved |
| Use DCA instead of PCA | Forecasting non-neural datasets · Temperature and Accelerometer | Future-state forecasting | Mixed |
| Use DCA instead of causal Kalman filtering | Latent-state extraction · M1 and Hippocampus | Behavioral-variable decoding held-out R² | Hurt |
| Use DCA instead of Kalman filtering or smoothing | Latent-state extraction · Temperature and Accelerometer | Forecasting held-out R² | Improved |
| Use DCA instead of causal GPFA | Latent-state extraction · M1 and Hippocampus | Behavioral-variable decoding held-out R² | Hurt |
| Use DCA instead of GPFA | Latent-state extraction · Temperature and Accelerometer | Forecasting held-out R² | Improved |
Closest papers by meaning
- Reduced-Rank Linear Dynamical Systems (AAAI 2018 · similarity 0.87)
- Learning Interpretable Hierarchical Dynamical Systems Models from Time Series Data (ICLR 2025 · similarity 0.86)
- Time-Series Decomposition as a Standalone Task: A Mechanism-Driven Diagnostic Benchmark (ICML 2026 · similarity 0.85)
- Learning Low-Dimensional Temporal Representations with Latent Alignments (TPAMI 2019 · similarity 0.85)
- Learning Low-Dimensional Temporal Representations (ICML 2018 · similarity 0.85)
- Learning Low-dimensional Latent Dynamics from High-dimensional Observations: Non-asymptotics and Lower Bounds (ICML 2024 · similarity 0.83)
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.
