We propose to train bi-directional neural network language model(NNLM) with noise contrastive estimation(NCE). Experiments are conducted on a rescore task on the PTB data set. It is shown that NCE-trained bi-directional NNLM outperformed the one trained by conventional maximum likelihood training. But still(regretfully), it did not out-perform the baseline uni-directional NNLM.
@article{arxiv.1602.06064,
title = {On Training Bi-directional Neural Network Language Model with Noise Contrastive Estimation},
author = {Tianxing He and Yu Zhang and Jasha Droppo and Kai Yu},
journal= {arXiv preprint arXiv:1602.06064},
year = {2016}
}