Residual networks classify inputs based on their neural transient dynamics

no code implementations • ICLR 2019 • Fereshteh Lagzi

We analyze the input-output behavior of residual networks from a dynamical system point of view by disentangling the residual dynamics from the output activities before the classification stage.

Compact representations and pruning in residual networks

no code implementations • 20 Oct 2018 • Fereshteh Lagzi, Tonio Ball, Joschka Boedecker

This criterion is based on the convergence of the neural dynamics in the last two successive layers of the residual block.