Stochastic Attention Head Removal: A Simple and Effective Method for Improving Transformer Based ASR Models

Abstract

Recently, Transformer based models have shown competitive automatic speech recognition (ASR) performance. One key factor in the success of these models is the multi-head attention mechanism. However, for trained models, we have previously observed that many attention matrices are close to diagonal, indicating the redundancy of the corresponding attention heads. We have also found that some architectures with reduced numbers of attention heads have better performance. Since the search for the best structure is time prohibitive, we propose to randomly remove attention heads during training and keep all attention heads at test time, thus the final model is an ensemble of models with different architectures. The proposed method also forces each head independently learn the most useful patterns. We apply the proposed method to train Transformer based and Convolution-augmented Transformer (Conformer) based ASR models. Our method gives consistent performance gains over strong baselines on the Wall Street Journal, AISHELL, Switchboard and AMI datasets. To the best of our knowledge, we have achieved state-of-the-art end-to-end Transformer based model performance on Switchboard and AMI.

Multi-Stream Acoustic Modelling Using Raw Real and Imaginary Parts of the Fourier Transform

“…We also compared the proposed models with other alternative systems including the end-to-end (E2E) stochastic attention head removal (SAHR) [70], multi-stream E2E [71] and the hybrid multi-scale octave CNNs [72], parametric (Parznet) 2- [73] 24.9 26.0 --Parznet 2D-CNN+VI (Hybrid) [12] 24. D CNNs without [73] and with variational inference (VI) [12].…”

Layer-wise Pruning of Transformer Attention Heads for Efficient Language Modeling

“…The scaling factor s g compensates for the masked portion and maintains the statistics after gating is applied. Recently, attention head dropout [17,18] has been introduced with similar scaling, but their scaling is used for regularization during training. We found that this technique greatly stabilizes the training dynamics, especially after the training is stabilized by the above two methods.…”