Research map / Time series / Foundation models for time series
TS2Vec: Towards Universal Representation of Time Series
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 TS2Vec instead of T-Loss | Unsupervised time-series classification · 125 UCR datasets | Avg. Acc.: 0.806 -> 0.830 (+2.4%) | Helped |
| Use TS2Vec instead of TS-TCC | Unsupervised time-series classification · 29 UEA datasets | Avg. Acc.: 0.682 -> 0.712 (+3.0%) | Helped |
| Use TS2Vec representations with linear regression | Univariate forecasting · Four forecasting datasets and horizons | Avg. MSE: 0.310 -> 0.209 | Helped |
| Use TS2Vec anomaly scoring | Univariate anomaly detection · Yahoo | F1: 0.563 -> 0.745 | Helped |
| Use TS2Vec anomaly scoring | Univariate anomaly detection · KPI | F1: 0.622 -> 0.677 | Helped |
| Add temporal contrast (ablation) | TS2Vec · 128 UCR datasets | Avg. Accuracy: -1.0% | Helped |
| Add instance contrast (ablation) | TS2Vec · 128 UCR datasets | Avg. Accuracy: -0.5% | Helped |
| Add hierarchical contrast (ablation) | TS2Vec · 128 UCR datasets | Avg. Accuracy: -1.7% | Helped |
| Add random cropping (ablation) | TS2Vec · 128 UCR datasets | Avg. Accuracy: -2.1% | Helped |
| Add timestamp masking (ablation) | TS2Vec · 128 UCR datasets | Avg. Accuracy: -0.9% | Helped |
| Add the input projection layer (ablation) | TS2Vec · 128 UCR datasets | Avg. Accuracy: -1.2% | Helped |
| Use contextual consistency instead of temporal consistency (ablation) | TS2Vec · 128 UCR datasets | Avg. Accuracy: -2.2% | Helped |
Closest papers by meaning
- T-Rep: Representation Learning for Time Series using Time-Embeddings (ICLR 2024 · similarity 0.90)
- TimesURL: Self-Supervised Contrastive Learning for Universal Time Series Representation Learning (AAAI 2024 · similarity 0.87)
- TSLANet: Rethinking Transformers for Time Series Representation Learning (ICML 2024 · similarity 0.87)
- TimeMixer++: A General Time Series Pattern Machine for Universal Predictive Analysis (ICLR 2025 · similarity 0.87)
- Exploring Representations and Interventions in Time Series Foundation Models (ICML 2025 · similarity 0.86)
- TS-RAG: Retrieval-Augmented Generation based Time Series Foundation Models are Stronger Zero-Shot Forecaster (NeurIPS 2025 · similarity 0.86)
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.