Filip Thor

This paper introduces a contrastive learning framework for dimensionality reduction of genetic data, showing improved preservation of global structure and generalization over existing methods.