About the Lab

Key Words: Machine Learning, Intelligent Data Processing, Voice/speech Processing, Computer Vision, Natural Language Processing, Data Science, etc.

How amazing it would be if machines could utilise their own "knowledge" to communicate with our human beings freely , creating various "interesting things" (e.g., art pictures, songs, etc.) with their own "abilities". However, the realisation of these technologies is challenging because it is quite different from the research methodology of modern science.

 The modern research methodology generally involves the objective observation, description, calculation and analysis of events from a human perspective, leading to the discovery of scientific knowledge. However, to make machines "can do" the intelligent work, such as the acquisition of knowledge and competence, it is completely opposite to the flow of modern research methodology - we have to let machines actively "learn" the knowledge from perceived information and subjectively generate experience and competence. In other words, we have to let the machine independently do everything on its own. Therefore, "learning" knowledge, "thinking", "deciding" an adaptive action in response to the complex situations surrounding the machine, and even freely "creating" things, such functions are really difficult to be implemented on the computer program.

 However, why is it necessary to realise human capabilities such as learning, understanding and creation on machines? Although the communication among human beings involves neurological and other actions, the most important one is considered to be the action of the "mind". When you realise intelligence in a machine, you are trying to instil a warm "heart" into a cold metal object. By such doing, we can get warm support and generous service from the machines, which will develop more business models. These are fascinating, which is why the matter of making machines intelligent is such significant.

 Now then, what kind of study is needed to instil the human "mind" into machines? In general, a wide range of knowledge seems required, and the Artificial Intelligence (AI) technology can comprehensively integrate these technologies. The AI is the programmed product of complex mathematical models, which has the ability to find patterns in the data that humans do not notice using machine learning processes. A successful AI model can turn the phenomena into "knowledge" (models), which can be improved by more effective mathematical theory or engineering approaches. Such technology makes it possible for machine devices to acquire "intellectual capacity" by learning from perceived data. Recently, technologies such as deep learning (e.g. Convolutional Neural Networks (CNNs), Generative Neural Networks (GANs), etc.), which mimic the human nervous system, have emerged along this direction.

 Our laboratory has been conducting research on the aforementioned topics for a long time. This lab is working on intelligent data processing, where we consider the intelligent capabilities can be "learned/mined" from the perceptual data. To make computers do such, we have been working on a wide range of issues related to computer vision, voice processing, natural language processing, machine learning, etc., which involve those essential capabilities of intelligent data processing. We welcome students undertaking the related research/ with a high interest in this area. If you consider to join us, should feel free to contact us.

(05/04/2022 updated, would continue to update)