Learning with Dynamic Architectures for Artificial Neural Networks - Adaptive Batch Size Approach

In this research we explore the performance of ADANET framework by using custom search space for an image-classification dataset using tensorflow libraries in combination with adaptive batch sizes for learning. In one experiment we classified fashion MNISET data and MNIST data of handwritten digits and obtained favorable results in terms of training time as well as accuracy by alternating learning batch sizes dynamically. Our testing was applied using simple deep neural network (DNN) and also with convolutional neural network (CNN).