中国科学院自动化研究所机构知识库

The conventional recipe for Automatic Speech Recognition (ASR) models is to 1) train multiple checkpoints on a training set while relying on a validation set to prevent over fitting using early stopping and 2) average several last checkpoints or that of the lowest validation losses to obtain the final model. In this paper, we rethink and update the early stopping and checkpoint averaging from the perspective of the bias-variance tradeoff. Theoretically, the bias and variance represent the fitness and variability of a model and the tradeoff of them determines the overall generalization error. But, it’s impractical to evaluate them precisely. As an alternative, we take the training loss and validation loss as proxies of bias and variance and guide the early stopping and checkpoint averaging using their tradeoff, namely an Approximated Bias-Variance Tradeoff  ApproBiVT). When evaluating with advanced ASR models, our recipe provides 2.5%–3.7% and 3.1%–4.6% CER reduction on the AISHELL-1 and AISHELL-2, respectively (The code and sampled unaugmented training sets used in this paper will be public available on GitHub).