Comparison of outlier detection methods on astronomical image data

TL;DR

This paper compares six unsupervised outlier detection methods applied to astronomical image data from SDSS stripe 82, evaluating their effectiveness in identifying rare objects and combining results for improved detection.

Contribution

It provides a systematic comparison of six outlier detection algorithms on astronomical data and proposes a method to combine their results for better outlier identification.

Findings

Different methods show varying sensitivity to hyperparameters.

Combining multiple methods improves outlier detection accuracy.

The study offers insights into the suitability of each method for astronomical data.

Abstract

Among the many challenges posed by the huge data volumes produced by the new generation of astronomical instruments there is also the search for rare and peculiar objects. Unsupervised outlier detection algorithms may provide a viable solution. In this work we compare the performances of six methods: the Local Outlier Factor, Isolation Forest, k-means clustering, a measure of novelty, and both a normal and a convolutional autoencoder. These methods were applied to data extracted from SDSS stripe 82. After discussing the sensitivity of each method to its own set of hyperparameters, we combine the results from each method to rank the objects and produce a final list of outliers.