A Novel Training Method for the Structured Language Frame Based on Neural Network

The Structured Language Frame aims m making a prediction of the next word in a given word string by making a syntactical analysis the preceding words. However faces the data sparseness problem because of the large dimensionality and diversity of the information available in the synthetic parses. In previous work [1, 2]. we proposed using neural network frames far the SLF. The neural network frame is better suited to tackle the data sparseness problem and its use gave significant improvements in perplexity and word error rate over the baseline SLF. In this paper we present a new method of the training the neural net based SLF. The presented procedure makes use of the partial parses hypothesized by the SLF itself, and is more expensive than the approximate trainig method used in previous work. Experiments with the new training method on the UPenn and WSJ corpora show significant reductions in perplexity and word error rate, achieving the lowest published results for the given corpora.