Efficient Neural Architecture Search for End-to-end Speech Recognition via Straight-Through Gradients

11 Nov 2020  ·  Huahuan Zheng, Keyu An, Zhijian Ou ·

Neural Architecture Search (NAS), the process of automating architecture engineering, is an appealing next step to advancing end-to-end Automatic Speech Recognition (ASR), replacing expert-designed networks with learned, task-specific architectures. In contrast to early computational-demanding NAS methods, recent gradient-based NAS methods, e.g., DARTS (Differentiable ARchiTecture Search), SNAS (Stochastic NAS) and ProxylessNAS, significantly improve the NAS efficiency. In this paper, we make two contributions. First, we rigorously develop an efficient NAS method via Straight-Through (ST) gradients, called ST-NAS. Basically, ST-NAS uses the loss from SNAS but uses ST to back-propagate gradients through discrete variables to optimize the loss, which is not revealed in ProxylessNAS. Using ST gradients to support sub-graph sampling is a core element to achieve efficient NAS beyond DARTS and SNAS. Second, we successfully apply ST-NAS to end-to-end ASR. Experiments over the widely benchmarked 80-hour WSJ and 300-hour Switchboard datasets show that the ST-NAS induced architectures significantly outperform the human-designed architecture across the two datasets. Strengths of ST-NAS such as architecture transferability and low computation cost in memory and time are also reported.

PDF Abstract


  Add Datasets introduced or used in this paper

Results from the Paper

Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Speech Recognition WSJ dev93 CTC-CRF ST-NAS Word Error Rate (WER) 5.68 # 1
Speech Recognition WSJ eval92 CTC-CRF ST-NAS Word Error Rate (WER) 2.77 # 5