High Order Recurrent Neural Networks for Acoustic Modelling
This work addresses a key bottleneck in training RNNs for speech recognition, offering a more parameter-efficient alternative to LSTM with competitive performance.
The paper tackled the vanishing gradient problem in recurrent neural networks by proposing a high order RNN (HORNN) with connections from multiple previous time steps, reducing word error rates by 4.2% and 6.3% over standard RNNs while using only 20%–50% of the parameters and computation compared to LSTM.
Vanishing long-term gradients are a major issue in training standard recurrent neural networks (RNNs), which can be alleviated by long short-term memory (LSTM) models with memory cells. However, the extra parameters associated with the memory cells mean an LSTM layer has four times as many parameters as an RNN with the same hidden vector size. This paper addresses the vanishing gradient problem using a high order RNN (HORNN) which has additional connections from multiple previous time steps. Speech recognition experiments using British English multi-genre broadcast (MGB3) data showed that the proposed HORNN architectures for rectified linear unit and sigmoid activation functions reduced word error rates (WER) by 4.2% and 6.3% over the corresponding RNNs, and gave similar WERs to a (projected) LSTM while using only 20%--50% of the recurrent layer parameters and computation.