Research map / Time series / Foundation models for time series
Finding Order in Chaos: A Novel Data Augmentation Method for Time Series in Contrastive Learning
Berken Utku Demirel, Christian Holz
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 proposed phase/magnitude mixup instead of linear mixup (ablation) | Ours (tailored mixup + SimCLR) · UCIHAR | accuracy: 83.09 -> 91.60 | Improved |
| use proposed phase/magnitude mixup instead of linear mixup (ablation) | Ours (tailored mixup + SimCLR) · USC | accuracy: 45.10 -> 60.13 | Improved |
| use proposed phase/magnitude mixup instead of linear mixup (ablation) | Ours (tailored mixup + SimCLR) · DaLia | MAE: 24.11 -> 10.57 | Improved |
| use proposed phase/magnitude mixup instead of linear mixup (ablation) | Ours (tailored mixup + SimCLR) · IEEE SPC12 | MAE: 20.45 -> 16.26 | Improved |
| use proposed phase/magnitude mixup instead of linear mixup (ablation) | Ours (tailored mixup + SimCLR) · Chapman | AUC: 86.30 -> 95.90 | Improved |
| sample mixing coefficients by latent-space distance (ablation) | Ours (tailored mixup + SimCLR) · UCIHAR | accuracy: 80.86 -> 91.60 | Improved |
| sample mixing coefficients by latent-space distance (ablation) | Ours (tailored mixup + SimCLR) · HHAR | accuracy: 87.53 -> 88.05 | Improved |
| sample mixing coefficients by latent-space distance (ablation) | Ours (tailored mixup + SimCLR) · IEEE SPC22 | MAE: 16.01 -> 12.25 | Improved |
| sample mixing coefficients by latent-space distance (ablation) | Ours (tailored mixup + SimCLR) · DaLia | MAE: 11.10 -> 10.57 (MAE); RMSE 20.13 -> 20.37 | Mixed |
| sample mixing coefficients by latent-space distance (ablation) | Ours (tailored mixup + SimCLR) · Chapman | AUC: 95.98 -> 95.90 | No clear effect |
| sample mixing coefficients by latent-space distance (ablation) | Ours (tailored mixup + SimCLR) · CPSC2018 | AUC: 80.67 -> 85.30 | Improved |
| use tailored mixup in contrastive learning instead of DACL hidden-representation mixup | SimCLR · UCIHAR | accuracy: 73.12 -> 91.60 | Improved |
Closest papers by meaning
- Parametric Augmentation for Time Series Contrastive Learning (ICLR 2024 · similarity 0.89)
- Frequency-Masked Embedding Inference: A Non-Contrastive Approach for Time Series Representation Learning (AAAI 2025 · similarity 0.85)
- SimPSI: A Simple Strategy to Preserve Spectral Information in Time Series Data Augmentation (AAAI 2024 · similarity 0.85)
- Time Series Generation Under Data Scarcity: A Unified Generative Modeling Approach (NeurIPS 2025 · similarity 0.84)
- Soft Contrastive Learning for Time Series (ICLR 2024 · similarity 0.84)
- An Unsupervised Approach for Periodic Source Detection in Time Series (ICML 2024 · 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.
