Progressive Neural Architecture Search
This work addresses the computational bottleneck in neural architecture search for researchers and practitioners, offering a more efficient method that is incremental but provides substantial speed improvements.
The paper tackles the problem of efficiently learning convolutional neural network structures, achieving up to 5 times greater efficiency in model evaluations and 8 times faster compute compared to prior reinforcement learning methods, while discovering architectures that achieve state-of-the-art classification accuracies on CIFAR-10 and ImageNet.
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.