CaptorX: A Class-Adaptive Convolutional Neural Network Reconfiguration Framework

Zhuwei Qin, Fuxun Yu, Zirui Xu, Chenchen Liu, Xiang Chen

Abstract: Nowadays, the evolution of deep learning and cloud service significantly promotes neural network-based mobile applications. Although intelligent and prolific, those applications still lack certain flexibility: for classification tasks, neural networks are generally trained with vast classification targets to cover various utilization contexts. However, only partial classes are practically inferred due to individual mobile user preference and application specificity, which causes unnecessary computation consumption. Thus, we proposed <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">CaptorX</i> —a class-adaptive convolutional neural network (CNN) reconfiguration framework to adaptively prune convolutional filters associated with unneeded classes. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">CaptorX</i> can reconfigure a pretrained full-class CNN model into class-specific lightweight models based on the visualization analysis of convolutional filters’ exclusive functionality for a single class. These lightweight models can be directly deployed to mobile devices without the retraining cost of traditional pruning-based reconfiguration. Furthermore, we can apply the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">CaptorX</i> framework into a distributed collaboration setting. With dedicated local training regulation and collaborative aggregation schemes, the class-adaptive models on individual mobile devices can further contribute back to the central full-class model. Experiments on representative CNNs and image classification datasets show that, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">CaptorX</i> can reduce the CNN computation workload up to 50.22% and save 46.58% energy consumption for varied local devices, meanwhile improving accuracy for their targeted classes with better task focus. With our distributed collaboration paradigm, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">CaptorX</i> also provides further potential to enhance the central model accuracy, while reducing up to 37.58% communication cost compared to traditional distributed learning methods.