arXiv 2411.02109
One protein is all you need
By Anton Bushuiev, Roman Bushuiev, et al.
Published 2024-11-04
Wiki summary
Explore the paper's summary, context, and related research on Papiers.
Generalization beyond training data remains a central challenge in machine learning for biology. A common way to enhance generalization is self-supervised pre-training on large datasets. However, aiming to perform well on all possible proteins can limit a model's capacity to excel on any specific one, whereas experimentalists typically need accurate predictions for individual proteins they study, often not covered i…