In this work, we investigate how to improve semi-supervised DNN for low resource languages where the initial systems may have high error rate. We propose using semi-supervised MLP features for DNN training, and we also explore using confidence to improve semi-supervised cross entropy and sequence training. The work conducted in this paper was evaluated under the IARPA Babel program for the keyword spotting tasks. We focus on the limited condition where there are around 10 hours of supervised data for training.
Bibliographic reference. Hsiao, Roger / Ng, Tim / Zhang, Le / Ranjan, Shivesh / Tsakalidis, Stavros / Nguyen, Long / Schwartz, Richard (2014): "Improving semi-supervised deep neural network for keyword search in low resource languages", In INTERSPEECH-2014, 1088-1091.