Activation Ensembles for Deep Neural Networks
This addresses the trial-and-error process in activation function selection for deep learning practitioners, though it appears incremental as it builds on existing activation function research.
The paper tackles the problem of selecting activation functions in neural networks by proposing activation ensembles, which allow multiple activation functions per layer and automatically select the most effective ones, achieving superior results on various datasets with feed-forward and convolutional networks.
Many activation functions have been proposed in the past, but selecting an adequate one requires trial and error. We propose a new methodology of designing activation functions within a neural network at each layer. We call this technique an "activation ensemble" because it allows the use of multiple activation functions at each layer. This is done by introducing additional variables, $α$, at each activation layer of a network to allow for multiple activation functions to be active at each neuron. By design, activations with larger $α$ values at a neuron is equivalent to having the largest magnitude. Hence, those higher magnitude activations are "chosen" by the network. We implement the activation ensembles on a variety of datasets using an array of Feed Forward and Convolutional Neural Networks. By using the activation ensemble, we achieve superior results compared to traditional techniques. In addition, because of the flexibility of this methodology, we more deeply explore activation functions and the features that they capture.