LATIDIA · Investigación

Las redes neuronales son más modulares de lo que sugieren las neuronas individuales

arXiv: 2610.12842v1Tipo de anuncio: nuevo Resumen: Las redes neuronales desarrollan rutinariamente representaciones en las que las neuronas individuales contribuyen a múltiples funciones distintas, lo que lleva a la opinión generalizada de que tales redes l

La noticia

arXiv:2610.12842v1 Announce Type: new Abstract: Neural networks routinely develop representations in which individual neurons contribute to multiple distinct functions, leading to the widespread view that such networks lack modular organization. We challenge this conclusion by introducing a formally nested hierarchy of modularity: unit-level, which requires distinct computations to occupy disjoint neurons; orthogonal, which asks whether a rotation reveals functionally independent subspaces; and linear, which asks whether a bounded linear map can separate task-relevant computations into independent subspaces. We develop a unified framework for learning decompositions at each level and evaluating them causally through selective ablation. Across vision models, language models and multi-task Recurrent Neural Networks (RNNs), we find that networks