Featureless Classification of Light Curves

TL;DR

This paper introduces a novel density model-based approach for classifying irregularly sampled light curves in astronomy, capturing comprehensive information and performing comparably to traditional feature-based methods.

Contribution

It proposes a density model representation for time series that preserves all available information, offering a principled alternative to feature extraction for classification tasks.

Findings

Performs comparably to state-of-the-art feature-based methods

Preserves all static information in observational data

Potential applicability to unsupervised learning tasks

Abstract

In the era of rapidly increasing amounts of time series data, classification of variable objects has become the main objective of time-domain astronomy. Classification of irregularly sampled time series is particularly difficult because the data cannot be represented naturally as a vector which can be directly fed into a classifier. In the literature, various statistical features serve as vector representations. In this work, we represent time series by a density model. The density model captures all the information available, including measurement errors. Hence, we view this model as a generalisation to the static features which directly can be derived, e.g., as moments from the density. Similarity between each pair of time series is quantified by the distance between their respective models. Classification is performed on the obtained distance matrix. In the numerical experiments, we use data from the OGLE and ASAS surveys and demonstrate that the proposed representation performs up to par with the best cur- rently used feature-based approaches. The density representation preserves all static information present in the observational data, in contrast to a less complete description by features. The density representation is an upper boundary in terms of information made available to the classifier. Consequently, the predictive power of the proposed classification depends on the choice of similarity measure and classifier, only. Due to its principled nature, we advocate that this new approach of representing time series has potential in tasks beyond classification, e.g., unsupervised learning.