Multi-objective Neural Architecture Search with Almost No Training
This work addresses the computational cost bottleneck of Neural Architecture Search (NAS) for researchers and practitioners, making NAS more accessible.
The authors propose Random-Weight Evaluation (RWE) to rapidly estimate neural network architecture performance by only training the last linear classification layer. This method reduces evaluation time from hours to seconds and, when integrated into a multi-objective evolutionary algorithm, finds state-of-the-art architectures on CIFAR-10 in under two hours on a single GPU.
In the recent past, neural architecture search (NAS) has attracted increasing attention from both academia and industries. Despite the steady stream of impressive empirical results, most existing NAS algorithms are computationally prohibitive to execute due to the costly iterations of stochastic gradient descent (SGD) training. In this work, we propose an effective alternative, dubbed Random-Weight Evaluation (RWE), to rapidly estimate the performance of network architectures. By just training the last linear classification layer, RWE reduces the computational cost of evaluating an architecture from hours to seconds. When integrated within an evolutionary multi-objective algorithm, RWE obtains a set of efficient architectures with state-of-the-art performance on CIFAR-10 with less than two hours' searching on a single GPU card. Ablation studies on rank-order correlations and transfer learning experiments to ImageNet have further validated the effectiveness of RWE.