2 papers
cs.LG2024
Controlling Grokking with Nonlinearity and Data Symmetry
Ahmed Salah, David Yevick
This paper demonstrates that grokking behavior in modular arithmetic with a modulus P in a neural network can be controlled by modifying the profile of the activation function as w…
cs.LG2024
Branched Variational Autoencoder Classifiers
Ahmed Salah, David Yevick
This paper introduces a modified variational autoencoder (VAEs) that contains an additional neural network branch. The resulting branched VAE (BVAE) contributes a classification co…