Abstract:
The past decade or so saw leap-frog progress of AI study. Some sectors, notably computer vision, data digging and information retrieval, show qualitative surge with new functions, commodities or even sectors popping up. It is noteworthy to mention that successful cases only linger on the assignment with so-called pattern recognition as main body according to extensive reckoning as regards AI setting and detailed application. Pattern recognition hereby refers to the assignment of recognizing rules and pattern concealed inside data or input and reflection of data to the knowledge structure through functions with neural network as mapping.
On the other hand, we hold that the next summit of AI is to work out machine decision-making, or feedback of decision to data before imposing an impact on it and forming a closed loop after acquiring knowledge other than data and mapping of knowledge. I am to focus on machine decision-making and enunciate how machines learn in the perplexed decision-making space. An AI system alone stops short of sating actual needs. In the present society, we have captured multiple agents driven by NN. I am to introduce our studies on how to dig into interaction and cooperation of multiple agents in both schools of the industrial world and how to become an indispensable topic for universal artificial intelligence.
Personal introduction:
Wang Jun, professor in the Department of Computer in UCL, Turing Fellow and guest professor in Alan Turing Research Institute, Shanghai Jiaotong University and Shanghai Science and Technology University. He is now on the academic furlough and is chief scientist for decision-making and reasoning in Noah Lab of Huawei. His major study is on intelligent information system including machine learning, intensified learning, multi-agent, data digging, computer advertising, recommended system and so forth. He has over 100 academic dissertations and two academic monographs in print and is a winner of Award for Best Dissertation for multiple times.