Searcharxiv⌕ Search

arXiv · 2609.36096

Comparing Active Anomaly Detection Frameworks: From Light Curves to Images

Abstract

The increasing volume and complexity of data from modern astronomical surveys are making comprehensive visual inspection increasingly impractical, motivating automated methods for identifying rare and unusual objects. Active anomaly detection (A-AD) combines unsupervised anomaly detection with iterative user feedback, allowing searches to adapt to scientific interests while retaining sensitivity to unexpected objects. We systematically compare three A-AD frameworks, Astronomaly, Protege, and PineForest, across Zwicky Transient Facility light curves and Euclid galaxy images. In baseline experiments, Protege, PineForest, and Astronomaly recovered 83, 78, and 38 variable stars, respectively, within the first 100 light-curve objects inspected, while PineForest recovered 70 merging or morphologically disturbed galaxies within the first 100 Euclid images, compared with 29 and 19 for Astronomaly and Protege. Experiments using prior labels show that their effectiveness depends on the composition of the prior sample and varies between A-AD methods. Finally, we combine the complementary behaviour of Protege and PineForest in a hybrid search of 10,000 Euclid images, recovering 799 morphologically unusual objects. These include 29 retained strong lens systems and candidates, of which 9 candidates have no counterparts in the published Euclid Q1 strong lens catalogues searched, and 202 visually classified ring galaxies. These results highlight the importance of selecting A-AD strategies according to both the scientific objective and dataset properties, and demonstrate their potential for efficient, discovery-driven exploration of forthcoming astronomical surveys.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Sreevarsha Sreejith, Robert C. Nichol. 2026-09-28. Comparing Active Anomaly Detection Frameworks: From Light Curves to Images. https://searcharxiv.com/abs/2609.36096

Citations retain the original authors and DOI and include this site’s record URL. If this discovery service contributed to your research, you can also cite Searcharxiv by Indiciae: BibTeX · RIS.

KEEP EXPLORING

Related papers

Optimal robust detection statistics for pulsar timing arrays

Pulsar timing arrays (PTAs) seek to detect a nano-Hz stochastic gravitational-wave background (GWB) by searching for the characteristic Hellings and Downs angular pattern of timing residual correlations. So far, the evidence remains below the conventional $5σ$ threshold, as assessed using the literature-standard "optimal cross-correlation detection statistic". While this quadratic combination of cross-correlated data maximizes the deflection (signal-to-noise ratio), it does not maximize the detection probability at fixed false-alarm probability (FAP), and therefore is not Neyman-Pearson (NP) optimal for the assumed noise and signal models. The NP-optimal detection statistic is a different quadratic form, but is not used because it also incorporates autocorrelations, making it more susceptible to uncertainties in the modeling of pulsar timing noise. Here, we derive the best compromise: a quadratic detection statistic which is as close as possible to the NP-optimal detection statistic (minimizing the variance of its difference with the NP statistic) subject to the constraint that it only uses cross-correlations, so that it is less affected by pulsar noise modeling errors. We study the performance of this new NPMV statistic for a simulated PTA whose noise and (putative) signal match those of the NANOGrav 15-year data release: GWB amplitude $A_{\rm gw} = 2.1 \times 10^{-15}$ and spectral index $γ= 13/3$. Compared to the literature-standard "optimal" cross-correlation detection statistic, the NPMV statistic increases the detection probability by 47% when operating at a $5σ$ false-alarm probability of $α= 2.9 \times 10^{-7}$.

astro-ph.IM↗

Lunar Reflective Interferometry

We present the method of lunar reflection interferometry (LRI) in which a virtual interferometer can be formed by using a spacecraft-borne antenna in lunar orbit that receives both direct rays and those reflected from the lunar surface. The technique exploits the method of images and is akin to the classic "sea cliff" interferometer, with the Moon's surface, primarily the lunar Maria, replacing the ocean surface. We describe the method in detail, demonstrate that significant portions of the Moon's surface are sufficiently smooth at low radio frequencies, $ν\lesssim 10\,\mathrm{MHz}$, for the technique to work, and outline a spacecraft instrument implementation. We describe potential systematic errors and how they could be mitigated. We present several astrophysics applications of the method.

astro-ph.IM↗

Wideband overlapped polyphase channelizer for kinetic inductance detector readout: split-path architecture and experimental characterization

Frequency-multiplexed readout of kinetic inductance detector arrays measures the complex transmission of thousands of superconducting resonators. The channelizer that separates them introduces systematic effects: a gain that depends on where each tone falls in the channel grid, leakage between channels and added noise. This article presents a 4096-channel polyphase channelizer with 2/1 overlap for a 4.096 GS/s input, implemented on a first-generation radio-frequency system-on-chip, and a procedure to characterize it on the board with two receive chains, one digitizing the excitation as a reference. A split-path architecture computes the two overlap states in parallel filter branches at 256 MHz and reorders them frame by frame, under three control rules; it keeps only the transform stage at 512 MHz and affords a prototype of 16 taps per polyphase path (65536 coefficients) at the multiplication rate of a published 8-tap bank. Each selected channel is exported as a complex series at 1 MS/s. The measured channel response follows the coefficient-table model within 0.15 dB down to -28 dB, with both overlapping channels at full gain across their common band. Leakage into every measured channel 1.875 MHz or more from the tone is -94 dBc or lower, apart from images attributed to converter interleaving. The signal-to-noise ratio per sample lies within 1.7 dB of the limit set by the data-sheet noise density of the converter, and an emulated detector pulse is recovered with a decay time of 14.65 $μ$s (standard uncertainty 0.04 $μ$s), within 0.7% of the expected value.

astro-ph.IM↗