Speaker
Description
Dwarf galaxies provide critical tests for cosmological models by probing $\Lambda$CDM predictions at sub-galactic scales. Despite their importance, detecting these faint systems beyond the Local Group remains challenging due to their diffuse nature and low surface brightness. In this talk, I will present results from our recent large-scale search for dwarf galaxy candidates in the Ultraviolet Near Infrared Optical Northern Survey (UNIONS) - a wide, deep, multi-band survey covering nearly 5,000 square degrees of the northern sky. We developed an automated detection pipeline that first preprocesses multi-band imaging data through binning, artifact removal, and stellar masking before employing the software MTObjects to detect low surface brightness objects. After parameter-based filtering and cross-matching between g, r, and i bands, we fine-tuned the deep learning model Zoobot, originally trained on Galaxy Zoo classifications, to identify dwarf candidates. Our training dataset incorporated visual classifications from multiple experts, capturing both consensus and uncertainty in dwarf identification. Applied to $\sim$1.5 million objects, our method identified over 23,000 dwarf galaxy candidates with probability scores > 0.8, of which $\sim$8,000 have probabilities exceeding 0.9. The spatial distribution of high-confidence candidates reveals a correlation with massive galaxies (log$(M_{*}/M_\odot) \geq$ 10) within 120$\,$Mpc, suggesting many are genuine satellites. Despite the high-confidence classification, these objects remain candidates that would benefit from spectroscopic follow-up to confirm their nature and obtain crucial distance measurements. We present this catalog as a community resource to advance studies of galaxy formation, hierarchical structure assembly, and the distribution of dwarf galaxies across diverse cosmic environments.
