Time Series Contrastive Learning

Time series contrastive learning aims to learn robust representations from sequential data by comparing augmented versions of time series. Current research focuses on improving the selection and generation of positive and negative pairs within a contrastive learning framework, addressing issues like false negatives, class imbalance, and the identification of "bad" pairs that hinder learning. This involves developing novel methods for data augmentation, often informed by information theory, and employing retrieval-based approaches or dynamic pair mining algorithms to enhance the quality of learned representations. These advancements are improving the performance of downstream tasks such as forecasting, classification, and anomaly detection across various time series modalities.

Papers

February 16, 2024

Parametric Augmentation for Time Series Contrastive Learning
Xu Zheng, Tianchun Wang, Wei Cheng, Aitian Ma, Haifeng Chen, Mo Sha, Dongsheng Luo
Contrastive Learning Time Series Time Series Representation Learning Time Series Contrastive Learning

December 19, 2023

Time-Series Contrastive Learning against False Negatives and Class Imbalance
Xiyuan Jin, Jing Wang, Lei Liu, Youfang Lin
Contrastive Learning Class Imbalance False Negative Instance Discrimination Time Series Contrastive Learning

November 1, 2023

REBAR: Retrieval-Based Reconstruction for Time-series Contrastive Learning
Maxwell A. Xu, Alexander Moreno, Hui Wei, Benjamin M. Marlin, James M. Rehg
Full State Reconstruction Contrastive Self Supervised Learning Contrastive Learning Framework Time Series Contrastive Learning Retrieval Based Reconstruction

June 11, 2023

Improving Time Series Encoding with Noise-Aware Self-Supervised Learning and an Efficient Encoder
Duy A. Nguyen, Trang H. Tran, Huy Hieu Pham, Phi Le Nguyen, Lam M. Nguyen
Robust Learning Contrastive Self Supervised Learning Time Series Representation Lightweight Encoder Encoder Architecture Informative Representation Time Series Contrastive Learning

March 21, 2023

Time Series Contrastive Learning with Information-Aware Augmentations
Dongsheng Luo, Wei Cheng, Yingheng Wang, Dongkuan Xu, Jingchao Ni, Wenchao Yu, Xuchao Zhang, Yanchi Liu, Yuncong Chen, Haifeng Chen, Xiang Zhang
Contrastive Learning Time Series Time Series Representation Learning Cut and Paste Time Series Contrastive Learning

February 7, 2023

Towards Enhancing Time Series Contrastive Learning: A Dynamic Bad Pair Mining Approach
Xiang Lan, Hanshu Yan, Shenda Hong, Mengling Feng
Contrastive Learning Time Series Representation Noisy Pair Hard Negative Mining Negative Pair Time Series Contrastive Learning

October 18, 2022

Universal hidden monotonic trend estimation with contrastive learning
Edouard Pineau, Sébastien Razakarivony, Mauricio Gonzalez, Anthony Schrapffer
Contrastive Learning Recent Trend Universal Image Time Series Contrastive Learning Monotonic Trend