Quality Alchemist (品質煉金術師): The 2nd Nation Higher Education Teacher Training Course for "Introduction to Artificial Intelligence" (第二期全国高校《人工智能导论》师资培训班) (Day 2)

2018年12月15日星期六

The 2nd Nation Higher Education Teacher Training Course for "Introduction to Artificial Intelligence" (第二期全国高校《人工智能导论》师资培训班) (Day 2)

The 2^nd Nation Higher Education Teacher Training Course for “Introduction of Artificial Intelligence” (第二期全国高校《人工智能导论》师资培训班) was organized by CAAI from 14^th to 16^th Dec 2018 in Fuzhou. In Day 2, I met Prof. Wang Wanliang (王万良) (浙江工业大学计算机科学与技术学院院长、软件学院院长、教授) and took a photo for memory. Prof. Wang Wanliang was old friend of Prof. Cai Wen (蔡文) and executive committee member of Extenics Society, Chinese Association for Artificial Intelligence (CAAI) (中國人工智能學會可拓學專業委員會).

Day 2 (20181215):

Prof. Wang Wanliang (王万良) was the first speaker in Day 2 for three subjects in the morning. Before the official topic, he introduced the “AI course development and education method” to us.

Firstly, Prof. Wang mentioned the “Higher Education for AI Innovation Action Plan” (高等学校人工智能创新行动计划) which was issued in April 2018. It included enhancing the AI innovation system, talent training and commercialization.

Then Prof. Wang compared AI talents distribution in the world. China was found much behind USA.

Finally, he explained some misunderstanding on AI education as follows:

Error 1: Poor background can’t learn it (基础差学不了)

Prof. Wong said University had already opened AI courses. High school and even primary school also taught AI courses.

Error 2: Studied but didn’t know how to do it. (学了不会做)

He explained the most important was taught the concept and student could think how to use AI to solve complex problem rather than solved by themselves. Thinking is much important than doing.

Error 3: Too theoretical and difficult to understand (太理论难理解)

He added that AI theories were not completed. But many things were similar human thinking to solve problem.

Error 4: Don’t know which books are suitable to read (不知道看什么书)

Prof. Wang told us that’s dependent on what purpose you selected.

Error 5: AI course difficult to teach (人工智能课程不好教)

Prof. Wang suggested to teach case study in application and development teaching project by topic, as well as, integrated research into teaching.

Then Prof. Wang started the first topic named “Knowledge Representation” (知识表示) that was the fundamental of human intelligent. It included knowledge concepts, generation methods, and framework, as well as, status space.

During the generation method, Prof. Wang briefed the three elements to express matter and affair. That’s same as “Extenics” such as Matter-element, Affair-element and Relation-element.

And then Prof. Wang mentioned the advantage and disadvantage of the generation method for knowledge representation.

The second topic entitled “Swarm Intelligence Algorithm” (群体智能算法). For instance, Swarm of bird to seek food was observed a higher efficient method than one bird.

After that Prof. Wang introduced “Genetic Algorithm” (遗传算法) through natural selection. It depended on “Crossover” (交叉) and 和“Mutation”(变异) .

He also introduced the process of the “Particle Swarm Optimization (PSO)” (粒子群優化算法) that was a computational method to optimize a problem by iteratively trying to improve a candidate solution with regard to a given measure of quality.

The last algorithm Prof. Wang briefed was “Ant Colony Optimization (ACO)” (蚂蚁算法) which was a probabilistic technique for solving computational problems to reduce for finding good paths through graphs.

The last topic by Prof. Wang in Day 2 morning was “Artificial Neural Network and Deep Learning” (人工神经网络与深度学习). In the beginning, Prof. Wang briefed natural neural network (NNN) and artificial neural network (ANN). Then he stated the history of ANN development.

And then he described the mathematical model of an artificial neuron (神经元數學模型).

After that he explained the Back Propagation (BP) Neural Network structure and its application to us.

Lastly, Prof. Wang introduced the “Generative Adversarial Network (GAN)” (生成对抗网络). He used one kind of Kung Fu in Martial arts novel named “左右互搏” as example to explain GAN that left hand fight with right hand for improvement.

The second speaker was Prof. Gao Yang (高阳) (南京大学计算机科学与技术系副主任; 人工智能教研室/实验室副主任、教授) and his topic entitled “The Importance of Multi-Agent Systems” (多智能体系統的重要性). Prof. Gao said the subject could be traced to the book “Perceptrons” in 1969 and “The Society of Mind” in 1986.

Then Prof. Gao briefed the history of Multi-Agent Systems from 40s-50s to 2002.

And then Prof. Gao introduced Autonomous Agent and Multi-Agent Systems (AAMAS) knowledge system that included Game Theory, Machine Learning, Sociology and Psychology. Based on AAMAS 2017, related papers submission in China had ~45, UK had ~63 and USA had ~129.

Prof. Gao also explained the Nash Equilibrium and Pareto Efficiency under game theory. After that he also introduced Plurality Protocol, Binary Protocol and Borad Protocol, as well as, different types of auction.

At the end, Prof. Gao introduced Reinforcement Learning (強化學習) in Multi-Agent Systems. There were three types: Cooperation, Competitive and Game.

The last speaker was Prof. Liu Hong (刘宏) (北京大学教授，中国人工智能学会副理事长) and his topic was “AI Technology in Robot Era” (机器人时代的人工智能技术). Prof. Liu Hong asked a question how long AI last to be hot.