Time series data is a crucial form of information that has vast opportunities. With the widespread use of sensor networks, largescale time series data has become ubiquitous. One of the most prominent problems in time series data mining is representation learning. Recently, with the introduction of self-supervised learning frameworks (SSL), numerous amounts of research have focused on designing an effective SSL for time series data. One of the current state-of-the-art SSL frameworks in time series is called TS2Vec. TS2Vec specially designs a hierarchical contrastive learning framework that uses loss-based training, which performs outstandingly against benchmark testing. However, the computational cost for TS2Vec is often significantly greater than other SSL frameworks. In this paper, we present a new self-supervised learning loss named, adaptive resolution loss. The proposed solution reduces the number of resolutions used for training the model via score functions, leading to an efficient adaptive resolution learning algorithm. The proposed method preserves the original model’s integrity while significantly enhancing its training time.
Garcia, Kevin, Juan Manuel Perez, and Yifeng Gao. 2023. “Adaptive Resolution Loss: An Efficient and Effective Loss for Time Series Self-Supervised Learning Framework.”