Shanghai Jiaotong University - Microsoft Intelligent Computing and Intelligent Systems Laboratory is currently the Microsoft Key Laboratory of the Ministry of Education, which established in September 2005 on the basis of many years of good cooperation between Shanghai Jiaotong University and Microsoft Research Asia. In order to better play in each of the concurrent computing, algorithms and complexity theory, imitation brain computing, computer vision, machine learning, computational advantages intelligence, natural language processing, multimedia communications and robotics fields, to achieve "future computers and robots capable of seeing, hearing, learning, natural language can communicate with humans, " the joint mission established. The laboratory also has achieved good results in scientific research, personnel training and academic exchanges, which cumulatively published more than 200 papers in top international conferences CVPR, ICCV, WWW and so forth.
Representative Results :
1 ) brain-machine interactive multi-modal driver fatigue detection system
The system by getting a driver's brain signals (EEG), eye signals (EOG), grip strength signals and Kinect image to extract the characteristics associated with fatigue from physiological signals and behavioral characteristics, create fatigue detection model using machine learning methods, achieve the measure and forecase of driver fatigue. Compared with traditional video-based fatigue detection methods , this system can improve the accuracy and reliability of the fatigue testing .
2 ) EEG-based brain-computer interaction and brain function rehabilitation training platform
The platform, using the brain-computer interface technology to read cortical limb movement intention ,obtaining neural feedback from the visual, auditory and tactile channels , and prompting the body to produce movement through functional electrical stimulation, thereby can establishs active limb movement control loop to improve motor function rehabilitation .
3 ) cognitive intelligent man-machine spoken dialogue system
Using a large vocabulary continuous speech recognition , synthesis parametric statistics , statistical semantic understanding , as well as context-based reasoning and dialogue and situational management techniques to construct the intelligent man-machine spoken dialogue system with real-time feedback, adaption and error correction capabilities.