DecisioNet: A Binary-Tree Structured Neural Network

Deep neural networks (DNNs) and decision trees (DTs) are both state-of-the-art classifiers. DNNs perform well due to their representational learning capabilities, while DTs computationally efficient as they inference along one route (root-to-leaf) that is dependent on the input data. In this paper, we present DecisioNet (DN), a binary-tree structured network. We propose systematic way convert an existing DNN into DN create lightweight version of original model. takes best worlds - it uses modules utilizes its tree structure only portion computations. evaluate various architectures, with corresponding baseline models FashionMNIST, CIFAR10, CIFAR100 datasets. show variants achieve similar accuracy significantly reducing computational cost