Time Series Classification and Analysis

The field of time series analysis is moving towards developing more efficient and effective methods for classification and clustering. Research is focused on addressing challenges such as domain shifts, irregular temporal data, and high-dimensional feature spaces. Innovative approaches include the use of optimal transport, Fisher information constraints, and soft sparse shape learning. These methods have shown promising results in improving the accuracy and robustness of time series classification models. Noteworthy papers include:

  • FIC-TSC, which proposes a training framework that leverages Fisher information as a constraint to enhance generalizability to distribution shifts.
  • PYRREGULAR, which introduces a unified framework for irregular time series classification and provides a standardized dataset repository.
  • Learning Soft Sparse Shapes, which proposes a soft sparse shape model for efficient time series classification.
  • 4TaStiC, which introduces a new clustering algorithm for long-term type 2 diabetes patients.
  • Topology-driven identification of repetitions in multi-variate time series, which presents a persistent homology framework to estimate recurrence times in multi-variate time series.

Sources

An Efficient Transport-Based Dissimilarity Measure for Time Series Classification under Warping Distortions

FIC-TSC: Learning Time Series Classification with Fisher Information Constraint

PYRREGULAR: A Unified Framework for Irregular Time Series, with Classification Benchmarks

Learning Soft Sparse Shapes for Efficient Time-Series Classification

4TaStiC: Time and trend traveling time series clustering for classifying long-term type 2 diabetes patients

Topology-driven identification of repetitions in multi-variate time series

Built with on top of