ApproBiVT: Lead ASR Models to Generalize Better Using Approximated Bias-Variance Tradeoff Guided Early Stopping and Checkpoint Averaging
This is an incremental improvement for ASR practitioners, offering a refined training recipe to enhance model performance.
The paper tackles the problem of improving generalization in automatic speech recognition models by proposing an approximated bias-variance tradeoff method for early stopping and checkpoint averaging, resulting in 2.5%-4.6% CER reduction on benchmark datasets.
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 overfitting 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.