Time-to-Pattern: Information-Theoretic Unsupervised Learning for Scalable Time Series Summarization

TL;DR

Time-to-Pattern (T2P) is a deep generative model that learns interpretable embeddings to identify diverse, salient patterns in time series data, improving scalability and pattern diversity over previous methods.

Contribution

The paper introduces T2P, a novel deep generative approach for unsupervised time series summarization that overcomes limitations of heuristic similarity measures and enhances pattern diversity.

Findings

T2P effectively discovers informative patterns in noisy, complex data.

T2P outperforms previous methods in pattern diversity.

T2P demonstrates improved scalability in processing large datasets.

Abstract

Data summarization is the process of generating interpretable and representative subsets from a dataset. Existing time series summarization approaches often search for recurring subsequences using a set of manually devised similarity functions to summarize the data. However, such approaches are fraught with limitations stemming from an exhaustive search coupled with a heuristic definition of series similarity. Such approaches affect the diversity and comprehensiveness of the generated data summaries. To mitigate these limitations, we introduce an approach to time series summarization, called Time-to-Pattern (T2P), which aims to find a set of diverse patterns that together encode the most salient information, following the notion of minimum description length. T2P is implemented as a deep generative model that learns informative embeddings of the discrete time series on a latent space specifically designed to be interpretable. Our synthetic and real-world experiments reveal that T2P discovers informative patterns, even in noisy and complex settings. Furthermore, our results also showcase the improved performance of T2P over previous work in pattern diversity and processing scalability, which conclusively demonstrate the algorithm's effectiveness for time series summarization.