
The early period of training significantly impacts out-of-distribution generalization in neural networks.


coremsg

early-period-of-training-impact-on-out-of-distribution-generalization


Early Period of Training Impact on Out-of-Distribution Generalization


title_rewrite


The training process of artificial neural networks can be interpreted as a dynamical system in graph space, exhibiting complex behaviors such as regular and chaotic dynamics depending on the learning rate regime.


dynamical-stability-and-chaos-in-artificial-neural-network-training-trajectories


Dynamical Stability and Chaos in Artificial Neural Network Training Trajectories
