- Attention Based Vision Mechanism There are huge quantities images and videos on the web.
A more effective indexing and retrieval approach for these resources is expected. This project is seeking a computational model for
visioin mechanism of attention in order to simulate the natural way to retrieve the interesting content
like human being by computer.
- Emotional Natural Language Understanding This project is seeking to categorize the natural language's emotional attribute and determine both the semantic orientation and the fine-grained emotions by a statistical method. For example, we analysis the emotion distribution in Chinese by a straightforward clustering method in sematic level.
- Facial Expression Processing The visual information during communication has always been neglected but
considerably useful for dialogue understanding in fact. In this project, we have been constructing a video based oriental
facial expression database during three years times and extracted the temporal variation as features. Next we will
employ different learning methods on these data in order to recognize the dynamic facial expression.
- NLP, Generation and Summarization The project was started by an automatic generation method in terms of key-words.