Online Sequence Training of Recurrent Neural Networks with Connectionist Temporal Classification

TL;DR

This paper introduces an online CTC training algorithm for RNNs that allows learning from longer sequences without extensive unrolling, enabling more efficient and scalable sequence modeling for speech recognition tasks.

Contribution

The authors propose an EM-based online CTC algorithm that supports training RNNs on sequences longer than the unrolling window, facilitating online learning and efficient GPU parallelization.

Findings

Achieved 20.7% PER on concatenated TIMIT sequences.

Trained WSJ speech model with only 64 unrollings, reducing memory usage.

Maintained competitive WER with significantly less unrolling.

Abstract

Connectionist temporal classification (CTC) based supervised sequence training of recurrent neural networks (RNNs) has shown great success in many machine learning areas including end-to-end speech and handwritten character recognition. For the CTC training, however, it is required to unroll (or unfold) the RNN by the length of an input sequence. This unrolling requires a lot of memory and hinders a small footprint implementation of online learning or adaptation. Furthermore, the length of training sequences is usually not uniform, which makes parallel training with multiple sequences inefficient on shared memory models such as graphics processing units (GPUs). In this work, we introduce an expectation-maximization (EM) based online CTC algorithm that enables unidirectional RNNs to learn sequences that are longer than the amount of unrolling. The RNNs can also be trained to process an infinitely long input sequence without pre-segmentation or external reset. Moreover, the proposed approach allows efficient parallel training on GPUs. For evaluation, phoneme recognition and end-to-end speech recognition examples are presented on the TIMIT and Wall Street Journal (WSJ) corpora, respectively. Our online model achieves 20.7% phoneme error rate (PER) on the very long input sequence that is generated by concatenating all 192 utterances in the TIMIT core test set. On WSJ, a network can be trained with only 64 times of unrolling while sacrificing 4.5% relative word error rate (WER).