ИСТИНА

The classification of volcanic structures on Venus remains a significant challenge in planetary science despite three decades of research following the Magellan mission (1990-1994). While previous studies have cataloged over 1,700 volcanic edifices, the distinction between coronae (annular tectonic features) and large volcanoes remains ambiguous due to subjective manual classification methods. Traditional approaches rely on topographic profiles or diameter-based criteria, but these methods lack objectivity and scalability for planetary-scale analysis. This study presents an automated machine learning framework combining variational autoencoders (VAEs) with clustering algorithms to: 1) Extract high-level features from SAR and topographic data; 2) Identify natural subclasses of volcanic structures; 3) Evaluate spatial distribution patterns of clustered features.