SpanSeq: similarity-based sequence data splitting method for improved development and assessment of deep learning projects

• DOI: 10.1093/nargab/lqae106
• Publication Date: 2024-08-16T10:09:54Z
• Authors: Alfred Ferrer Flo...; Jose Juan Almagro...; Henrik Nielsen; Frank Møller Aare...; Philip Thomas Lan...
• Keywords: Similarity (geometry); Biological data; Sequence (biology)

The use of deep learning models in computational biology has increased massively recent years, and it is expected to continue with the current advances fields such as Natural Language Processing. These models, although able draw complex relations between input target, are also inclined learn noisy deviations from pool data used during their development. In order assess performance on unseen (their capacity