Expediting Training Using Information Theory Based Patch Ordering Algorithm
Henok Ghebrechristos, Gita Alaghband Gita · 2018
We present a framework for automatically ordering image patches that enables in-depth analysis of dataset relationship to learnability of a classification task using convolutional neural network. Our preliminary experimental results show that an informed smart shuffling of patches at a sample level can expedite training by exposing important features at early stages of training. Using multiple network architectures and datasets, we show that ordering image regions using mutual information measure between adjacent patches, enables CNNs to converge in a third of the total steps required to train the same network without patch ordering.