AI System Enables Semantic Search of 100 Million Galaxy Images, Discovers New Stellar Streams
Researchers developed AION-Search, an AI system that uses Vision-Language Models to generate descriptions of unlabeled galaxy images and enable semantic searching across over 100 million images. The method combines AI-generated captions with a pre-trained astronomy foundation model to create searchable embeddings without manual labeling. The system achieved state-of-the-art performance in finding rare astronomical phenomena and identified 36 new extragalactic stellar stream candidates.
A new AI pipeline called AION-Search makes it possible to semantically search massive archives of unlabeled galaxy images by leveraging Vision-Language Models (VLMs) to automatically generate descriptions. The system works by generating captions for galaxy images, then contrastively aligning a pre-trained astronomy foundation model with these embedded descriptions to produce searchable embeddings at scale. Testing showed the method outperforms direct image similarity search and achieves state-of-the-art zero-shot performance in identifying rare astronomical phenomena, even though it was trained on randomly selected images with no deliberate curation for rare cases. The researchers also introduced a VLM-based re-ranking method that nearly doubles recall for challenging targets in top-100 results. The system has already enabled the identification of 36 new extragalactic stellar stream candidates. Beyond astronomy, the approach demonstrates a generalizable method for making large, unlabeled scientific image archives semantically searchable across fields including Earth observation and microscopy.
What's missing
The study does not discuss computational costs, inference time, or hardware requirements for processing 100M+ images. Additionally, the paper does not address potential failure modes or systematic biases that VLM-generated descriptions might introduce when applied to astronomical data, nor does it discuss validation methods for the 36 newly identified stellar stream candidates.
What different sources said
- arXiv cs.AICenter
Semantic search for 100M+ galaxy images using AI-generated captions
Related
Genetic Drift, Not Selection, Drives Rapid Feather Color Evolution in Island Bird Radiation
A new study of an island bird radiation found that rapid evolution of feather coloration is driven primarily by genetic drift in small populations rather than sexual or ecological selection. The research integrated whole-genome data with detailed plumage measurements across complete species sampling to test whether signaling trait evolution correlates with speciation rates. The findings suggest that neutral demographic processes play a central role in generating phenotypic diversity during island radiations, challenging assumptions about the mechanisms driving rapid evolution.
New AI Model Improves Prediction of Therapeutic Peptide Function from Protein Sequences
Researchers developed a lightweight CNN classifier that predicts whether peptide sequences have therapeutic properties, trained on a database of 54,655 peptides across 48 functional categories. The model uses a novel negative sampling strategy to reduce false positive rates from over 60% in previous approaches to 2.1%. This advancement could accelerate drug discovery by enabling faster computational screening of peptide candidates before expensive experimental testing.
Study Shows Different Metabolic Stress Models Produce Distinct Effects on Human Neuronal Networks
Researchers tested three common in vitro metabolic stress models on human-derived neuronal networks and found each produced different patterns of neuronal activity and cell damage. The models tested were hypoxia alone, oxygen-glucose deprivation (OGD), and hypoxia combined with glutamate exposure. The findings suggest that choice of experimental model significantly affects results and that combining electrophysiological and structural analyses is important for accurately assessing metabolic stress in stroke research.