Current browse context:
cs.CV
Change to browse by:
References & Citations
Computer Science > Computer Vision and Pattern Recognition
Title: Progressive Neural Architecture Search
(Submitted on 2 Dec 2017 (v1), revised 24 Mar 2018 (this version, v2), latest version 26 Jul 2018 (v3))
Abstract: We propose a new method for learning the structure of convolutional neural networks (CNNs) that is more efficient than recent state-of-the-art methods based on reinforcement learning and evolutionary algorithms. Our approach uses a sequential model-based optimization (SMBO) strategy, in which we search for structures in order of increasing complexity, while simultaneously learning a surrogate model to guide the search through structure space. Direct comparison under the same search space shows that our method is up to 5 times more efficient than the RL method of Zoph et al. (2018) in terms of number of models evaluated, and 8 times faster in terms of total compute. The structures we discover in this way achieve state of the art classification accuracies on CIFAR-10 and ImageNet.
Submission history
From: Chenxi Liu [view email][v1] Sat, 2 Dec 2017 06:23:16 GMT (750kb,D)
[v2] Sat, 24 Mar 2018 00:39:27 GMT (2037kb,D)
[v3] Thu, 26 Jul 2018 19:51:26 GMT (1539kb,D)
Link back to: arXiv, form interface, contact.