skip to main content
Idioma:

Practical Block-Wise Neural Network Architecture Generation

Zhong, Zhao ; Yan, Junjie ; Wu, Wei ; Shao, Jing ; Liu, Cheng-Lin

2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018, p.2423-2432

IEEE

Texto completo disponível

Citações Citado por
  • Título:
    Practical Block-Wise Neural Network Architecture Generation
  • Autor: Zhong, Zhao ; Yan, Junjie ; Wu, Wei ; Shao, Jing ; Liu, Cheng-Lin
  • Assuntos: Computer architecture ; Convolutional codes ; Convolutional neural networks ; Indexes ; Network architecture ; Task analysis
  • É parte de: 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018, p.2423-2432
  • Descrição: Convolutional neural networks have gained a remarkable success in computer vision. However, most usable 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 sequentially to choose component layers. 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 performs competitive results in comparison to the hand-crafted state-of-the-art networks on image classification, additionally, the best network generated by BlockQNN achieves 3.54% top-1 error rate on CIFAR-10 which beats all existing auto-generate networks. (2) in the meanwhile, it offers tremendous reduction of the search space in designing networks which only spends 3 days with 32 GPUs, and (3) moreover, it has strong generalizability that the network built on CIFAR also performs well on a larger-scale ImageNet dataset.
  • Editor: IEEE
  • Idioma: Inglês
Voltar para lista de resultados

  • Realização: ABCD-USP USP   Logos de Redes Sociais: Freepik

Buscando em bases de dados remotas. Favor aguardar.