LATIDIA · Ciberseguridad

LEAN: Modulación de ingeniería de pérdida del modelo de distribución de sensibilidad al peso para redes neuronales eficientes y tolerantes a errores

arXiv:2610.28872v1 Tipo de anuncio: cruz Resumen: Los pesos de una red entrenada no tienen una sensibilidad uniforme a la precisión del modelo: los pesos más sensibles del 1% conllevan aproximadamente una quinta parte del daño que puede hacer un poco de volteo. Thi

La noticia

arXiv:2610.28872v1 Announce Type: cross Abstract: The weights of a trained network do not have uniform sensitivity to model accuracy: the 1% most sensitive weights carry about a fifth of the damage a bit flip can do. This distribution decides the bit-flip tolerance of a model, the fraction of weights a code must cover and the bits each weight needs; existing schemes work with the distribution training leaves. We show that we can systematically modulate or alter the weight sensitivity distribution of a trained model, thereby shifting weights from one sensitivity bin to another. The sensitivity of a stored weight is the product of its gradient and the step a flip