BlockQNN: Efficient Block-Wise Neural Network Architecture Generation

• Published in: IEEE transactions on pattern analysis and machine intelligence Vol. 43; no. 7; pp. 2314 - 2328
• Main Authors: Zhong, Zhao, Yang, Zichen, Deng, Boyang, Yan, Junjie, Wu, Wei, Shao, Jing, Liu, Cheng-Lin
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.07.2021; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Acceleration; Artificial neural networks; AutoML; Computer architecture; Computer vision; Construction; Convolutional neural network; Graphics processing units; Image classification; Indexes; Machine learning; Network architecture; neural architecture search; Neural networks; Q-learning; reinforcement learning; Strategy; Task analysis
• ISSN: 0162-8828; 1939-3539; 2160-9292; 1939-3539
• DOI: 10.1109/TPAMI.2020.2969193

Convolutional neural networks have gained a remarkable success in computer vision. However, most popular network architectures are hand-crafted and usually require expertise and elaborate design. In this paper, we provide a block-wise network generation pipeline called BlockQNN which automatically builds high-performance networks using the Q-Learning paradigm with epsilon-greedy exploration strategy. The optimal network block is constructed by the learning agent which is trained to choose component layers sequentially. We stack the block to construct the whole auto-generated network. To accelerate the generation process, we also propose a distributed asynchronous framework and an early stop strategy. The block-wise generation brings unique advantages: (1) it yields state-of-the-art results in comparison to the hand-crafted networks on image classification, particularly, the best network generated by BlockQNN achieves 2.35 percent top-1 error rate on CIFAR-10. (2) it offers tremendous reduction of the search space in designing networks, spending only 3 days with 32 GPUs. A faster version can yield a comparable result with only 1 GPU in 20 hours. (3) it has strong generalizability in that the network built on CIFAR also performs well on the larger-scale dataset. The best network achieves very competitive accuracy of 82.0 percent top-1 and 96.0 percent top-5 on ImageNet.