
Introducing a technique to convert neural networks from weight space to function space through dual parameterization, enabling efficient retention of knowledge and incorporation of new data without retraining.


coremsg

function-space-parameterization-of-neural-networks-for-sequential-learning-a-novel-approach


Function-space Parameterization of Neural Networks for Sequential Learning: A Novel Approach


title_rewrite


Neural networks can be converted to function space through a dual parameterization, enabling efficient sequential learning.


neural-networks-function-space-parameterization-for-sequential-learning-at-iclr-2024


Neural Networks Function-space Parameterization for Sequential Learning at ICLR 2024
