1. Deep Learning for Language Understanding
For understanding languages including Korean, we develop distributed representation of words with deep/recurrent neural architectures. The representations, which is also called “neural embedding”, of each word, sentence, or documents can be used for synonym suggestion, automatic sentence completion, semantic document search, and document summarization.
2. Deep Learning for Speech Recognition
We aim to improve speech recognition system by providing new framework such as new neural network for AM, LM and end-to-end ASR without HMM.
3. Deep Learning for Computer Vision
Large-scale object recognition and facial expression analysis based on deep convolutional neural networks (deep CNNs)