Research map / Time series / Foundation models for time series
Dynamic Sparse Network for Time Series Classification: Learning What to “See”
Qiao Xiao, Boqian Wu, Yu Zhang, Shiwei Liu, Mykola Pechenizkiy, Elena Mocanu, Decebal Constantin Mocanu
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 sparse kernels (S=80%) instead of dense kernels (ablation) | DSN · EEG2 | accuracy (%): 89.60 -> 99.10 | Improved |
| use sparse kernels (S=80%) instead of dense kernels (ablation) | DSN · HAR | accuracy (%): 96.06 -> 96.82 | Improved |
| use sparse kernels (S=80%) instead of dense kernels (ablation) | DSN · Daily Sport | accuracy (%): 99.19 -> 99.19 | No clear effect |
| increase sparsity from 50% to 80% (ablation) | DSN · EEG2 | accuracy (%): 96.23 -> 99.10 | Improved |
| increase sparsity from 80% to 90% (ablation) | DSN · EEG2 | accuracy (%): 99.10 -> 98.90 | No clear effect |
| add kernel grouping to dynamic sparse training (ablation) | DSN · EEG2 | accuracy (%): 96.77 -> 97.87 (S=60%), 97.27 -> 98.70 (S=70%), 98.77 -> 99.10 (S=80%) | Improved |
| add kernel grouping to dynamic sparse training (ablation) | DSN · HAR | accuracy (%): 96.87 -> 96.53 (S=60%), 96.52 -> 97.14 (S=70%), 96.79 -> 96.82 (S=80%) | Mixed |
| add kernel grouping to dynamic sparse training (ablation) | DSN · Daily Sport | accuracy (%): 99.23 -> 99.27 (S=60%), 99.25 -> 99.20 (S=70%), 99.28 -> 99.19 (S=80%) | No clear effect |
| dynamic sparse training instead of fixed topology (DSN fix) (ablation) | DSN · UCR 85 archive | accuracy | Improved |
| grow weights by gradient (RigL-style) instead of randomly (ablation) | DSN · UCR 85 archive | accuracy | No clear effect |
| DSN versus OS-CNN, InceptionTime, ResNet | DSN · UCR 85 archive | datasets won: 53-29-3 vs ResNet; 40-40-5 vs InceptionTime; 49-33-3 vs OS-CNN | Improved |
| DSN versus MLSTM-FCN on Daily Sport | DSN · Daily Sport | accuracy (%): 99.65 vs 99.19 | Hurt |
Closest papers by meaning
- MPTSNet: Integrating Multiscale Periodic Local Patterns and Global Dependencies for Multivariate Time Series Classification (AAAI 2025 · similarity 0.86)
- TimeMIL: Advancing Multivariate Time Series Classification via a Time-aware Multiple Instance Learning (ICML 2024 · similarity 0.85)
- Advancing Dynamic Sparse Training by Exploring Optimization Opportunities (ICML 2024 · similarity 0.84)
- STanHop: Sparse Tandem Hopfield Model for Memory-Enhanced Time Series Prediction (ICLR 2024 · similarity 0.84)
- Navigating Extremes: Dynamic Sparsity in Large Output Spaces (NeurIPS 2024 · similarity 0.84)
- TVNet: A Novel Time Series Analysis Method Based on Dynamic Convolution and 3D-Variation (ICLR 2025 · 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.
