AI Industry Synthesis Report.pdf

Yiou IntelligenceAI Industry Synthesis ReportApril,2017AI Industry Synthesis Report亿欧智库:人工智能产业综述报告Machines can not only extend humans’ hands and feet, replacing humanson some manual labor, but also extend humans’ brains, replacing somehumans’ thinking activities. It has already been an obvious fact since theadvent of computers. Following this direction, a new technology of artificialintelligence (which we use AI as abbreviation in following) will be formedgradually in the future, with the development of computer technique.AI includes but is not limited in simulating the central nervous system inthe human’s brain to build neural network models to deal with complexand changing problems in the reality. It could instruct the computer to playchess,provetheorems,draftstrategiesandmakedecisions,usemachines to recognize the words, sound, and images, contact withcomputers directly through the natural language (human language), anddevelop machines with the capability of recognition, analysis, synthesisand execution, and so on.From the time human first created the computer which needs guiding andcontrolling to run, to the time when the computers could learn bythemselves, there is a qualitative leap for the academic community and itbrings economic benefits of the tangible productivity to the industry.We believe that 2016 is an important turning point for the development ofAI. It is faster and stronger computing power and the explosively growingdatabase that push AI to the spotlight. It is not too much to say that AI hasbeen at the forefront of popular culture and political discourse. However,we also speculate that it is also possible that AI will be at the cusp in thenext intersection.The purpose of this report is to review the route of the development of AI,to examine the nature of AI technology, to unearth the opportunities andchallenges for technology landing in the specific market with a researchingattitude, and help with the continuous development of AI before the comeof the time “giving people promises and disappointing them”.Introduction2Publisher：Yiou IntelligenceAuthor：Miaoyun WuSupervisor：Tianyu YouOther Support：Bowei Du, Chao Wang, Zhou Chen, Kaili Guo and othersTranslator：Zeyao Xiao，ShuangWangAI Industry Synthesis Report亿欧智库:人工智能产业综述报告CatalogueCHAPTER 1Reviewing the history of AI development1.1 Definition of AI1.2 Development of Neural Network1.3 The opportunities and challenges of Deep LearningCHAPTER 2Thinking about the nature of AI2.1 Overview of all kinds of AI technologies2.1 computer vision technology2.2 Speech Recognition Technology2.3 Natural Language Understanding TechnologyCHAPTER 3The opportunities and challenges of the application of AI in some specific areas3.1 Why we believe the application of AI will firstly happen on health care, finance and transportation3.2 AI+ health care3.3 AI+ finance3.4 AI+ transportationAppendixYiou Intelligence: list of the core AI talents from the main enterprises in China (215 people)3AI Industry Synthesis Report亿欧智库:人工智能产业综述报告CHAPTER 1Reviewing the history of AI development1.1 Definition of AIWhat is AI?Tracing the generation and development of AIAI, machine learning and deep learning1.2 Development of neural networkFrom neural module to deep learning1.3 The opportunities and challenges of deep learning4AI Industry Synthesis Report亿欧智库:人工智能产业综述报告人工智能发展路线图·What is AI?In general, AI is to create a machine which can think like humans. It not only exists in science fictionfilms, but also is the ultimate dream of science geeks, and the synonymous of the impractical,whimsical fantasy of the technology fans.For the narrow sense of AI, Professor Nielson from Stanford University’s AI Research Center in U.S.A.defined it as: "AI is the subject about knowledge, is the science about how to express knowledge andhow to acquire and use knowledge. Defining AI by the functions it achieved, it is the functionsperformed by intelligent machines which is usually human intelligence-related, such as judging,proving, identification learning and problem solving and other thinking activities. These reflect thebasic ideas and basic aspects of the AI discipline that AI is to study the rule of human intelligenceactivity. In a practical view, AI is a knowledge engineering: to have knowledge as the object, study theacquisition of knowledge, the method of performing knowledge and the use of knowledge. "·Tracing the generation and development AIThe earliest artificial neural model was proposed in 1943. A group of scientists discussed at theDartmouth Conference in 1956. They finally proposed the concept of AI and the year then becamethe first year of AI. In the Fifth International AI Conference in 1977, computer scientist ProfessorFergenbaum from Stanford University formally put forward the concept of knowledge engineering.Various expert systems were then developed, and many commercialized expert systems wereintroduced to the market. Unfortunately, the development did not last long due to the lack ...