Interpretation of AI behind the hottest intelligen

  • Detail

Interpretation of artificial intelligence behind intelligent manufacturing

intelligent manufacturing is the deep integration of advanced manufacturing technology and information technology. It is a common enabling technology for the transformation and upgrading of China's traditional industries and the development of strategic emerging industries. Intelligent manufacturing corresponds to the fourth industrial revolution. Among its nine technical pillars, industrial IOT, cloud computing and industrial big data are the Three Foundations Based on distribution and connection. Industrial robots and 3D printing are the two hardware tools, knowledge work automation and industrial network security are the two software supports, and virtual reality and artificial intelligence are the two traction technologies for the future

what is artificial intelligence

artificial intelligence is a kind of machine intelligence, a system or discipline in which machines simulate or simulate human intelligence. The main research contents of artificial intelligence include cognitive modeling, knowledge representation, reasoning and application, machine perception, machine thinking, machine learning, machine behavior and intelligent systems, etc. reasoning, knowledge, planning, learning, communication, perception, mobility, operation, etc. These are the basic things to be studied in artificial intelligence

when it comes to artificial intelligence, there is a gentleman we can't help mentioning - Alan Turing. He is the father of both computer and artificial intelligence. Turing test, a test used to judge whether a machine has intelligence, was proposed by him and named after him. In 1956, more than a dozen famous scientists gathered in Dartmouth and spent two months discussing the problem of artificial intelligence. Since then, a new interdisciplinary field - the field of artificial intelligence has been opened up for more than 60 years. AI has experienced ups and downs in the past 60 years

three waves of artificial intelligence

artificial intelligence has three sects. The first sect is usually called logicism, which is also called symbolism with the development of the new material industry supported by the state. The core is symbolic reasoning and machine reasoning, which uses symbolic expression to study intelligence and reasoning. The second sect is called connectionism. Its core is neuron networks and deep learning. It imitates human nervous system, presents the model of human nervous system in a computational way, and uses it to imitate intelligence. The third sect is behaviorism, which advocates control, adaptation and abandonment of use to re purchase and evolutionary computing, which is rarely mentioned at present

the first wave of artificial intelligence was from 1956 to 1976. Symbolism, machine proof and artificial intelligence logic language made rapid progress. At that time, the biggest achievements were expert systems and knowledge engineering. Artificial intelligence is very popular in the early stage of development and is widely valued. In 1958, two years after the birth of AI, two masters (Simon and Newell) made a famous prediction that AI could solve many things within 10 years. For example:

defeat the chess champion within 10 years

discover and prove meaningful mathematical theories within 10 years

write beautiful fun within 10 years

achieve most psychological theories within 10 years

but 20 years later, the ambitious goals set in knowledge engineering are mostly difficult to achieve. Minsky published the article "k-lines: a theory of memory" in 1979, Basically, the ability of large-scale learning of neural networks was denied, symbolism and connectionism were depressed, and artificial intelligence entered the first low tide and ice age

the 30 years from 1976 to 2006 are the second wave of artificial intelligence. In this wave, after several milestone works, connectionism has risen again. Among them, BP network proposed in 1986 proved for the first time that the learning and training process of neural network can converge, which can be said to be the foundation work of the whole wave of artificial intelligence

although the improvement of neural network theory has made connectionism popular again, until 2006, artificial intelligence was still unable to get out of the theoretical research of the laboratory and was difficult to be really applied to the industry. In 2006, Geoffrey Hinton, together with Yann Lecun and Joshua bengio, published a breakthrough paper, "a fast learning algorithm for deep belief nets", which theoretically solved the problem that the scale of the original neural network could not be expanded, It can only deal with single situations, but can not deal with complex situations, which directly promotes the breakthrough of deep learning theory, and has developed all the way to today's height, forming the third wave of artificial intelligence

in essence, there is no essential difference between the second wave and the third wave in methodology. The difference lies in the success of deep learning. The progress of hardware and convolution neural network model and parameter training skills are two important factors that contribute to the success of deep learning

current situation of artificial intelligence

with the arrival of the third wave of artificial intelligence, the research of using computers to simulate human thinking process and intelligent behavior has been greatly developed through "machine learning" and "deep learning". After the application of deep neural network in speech recognition and image recognition, the breakthrough is particularly obvious

take the field of intelligent voice interaction as an example, which is a core technology field with high threshold and fierce global competition. Before 2000, the Chinese voice industry was controlled by international IT giants such as Microsoft and IBM. Now, iFLYTEK in China has become the largest voice and artificial intelligence listed company in the Asia Pacific region. Major Internet giants at home and abroad, including Google, apple, Baidu, Tencent, Alibaba, have established their own voice R & D teams. In this era of mobile Internet, thousands of mobile Internet app applications have access to voice cloud platform services, and voice interactive services have been applied in more and more industries, including automotive, automotive and smart home appliances. In the field of aerospace, the national pillar industry, voice control and voice interaction functions have long been unique to foreign fighters. Because most thermoplastic materials will not be interrupted in this experiment, it will become a technology supplier and product manufacturer focusing on voice interaction solutions in the aviation field. The intelligent cockpit voice control module and products developed by Hangfei Technology Co., Ltd. have been installed and tested on many types of fighter aircraft in domestic mainframe institutes

in the past 2016, the performance of speech recognition has made breakthroughs. In February, the phrase recognition word error rate of Baidu deep speech2 engine fell to 3.7%; In May, the conversational word error rate of IBM Watson system was as low as 6.9%; In September, the word error rate of English speech recognition of Microsoft's new system was as low as 6.3%, and further reduced to 5.9% in October, comparable to human beings. According to the statistics of an experiment conducted by Stanford University, on mobile devices, whether in Chinese or English, the input speed of voice is three times faster than that of manual typing

in the field of intelligent manufacturing, industrial robots have become an outstanding representative of artificial intelligence and have been repeatedly mentioned as one of the important implementation ends of intelligent manufacturing. Many manufacturing enterprises explore and lead the development of intelligent robots in intelligent manufacturing, and cross-border cooperation inside and outside the industrial control circle is everywhere. In 2013, Google incorporated eight robotics companies including Boston Dynamics. In 2014, AI company vicarious received a $12million investment led by abbtechnology | ventures. In 2015, Alibaba and Foxconn made strategic investment in sbrh, a robotics company owned by Softbank of Japan. In 2016, Siemens' new business unit next47 was officially established to tap Siemens' disruptive innovative ideas and potential in the core business areas of electrification, automation and digitalization, and its focus includes artificial intelligence. At the 2016 Industrial Expo, FANUC intelligent edge link and drive (field) system was jointly released by FANUC, Cisco and Rockwell Automation. Fieldsystem realizes advanced machine learning and deep learning capabilities, and combines artificial intelligence and cutting-edge computer technology to make distributed learning possible. The operation data of robots and equipment are processed in real time on the network, which also makes the coordinated production between various equipment more intelligent, and makes the complex production coordination that was difficult to achieve possible

liuqingfeng, chairman of iFLYTEK, proposed to take AI strategy as a national key area at the two sessions this year. He believed that even if there is no major technological breakthrough in the next three to five years, with the existing AI technology elements, a huge application breakthrough and industrial breakthrough will be formed. Which country gives priority to occupy the commanding height, artificial intelligence will be in the forefront of the world in the future

the development of artificial intelligence has experienced two ups and downs, and now it is in a good period of the third vigorous rise. At present, it is not advisable to exaggerate and belittle artificial intelligence. Many research results of artificial intelligence can be expected to be really used in the production process of manufacturing in a few years. But today, it still needs time to optimize and mature its usability

the intelligent manufacturing we want to realize now is intelligent manufacturing based on artificial intelligence, not just artificial intelligence. Among them, industrial intelligence derived from the long-term accumulation in the industrial field and artificial intelligence derived from the information field need to learn from and integrate with each other. Taking these two intelligent technologies as the main body and taking into account other intelligent technologies is the mainstream development direction of intelligent manufacturing technology in the future. Starting from industrial intelligence and gradually integrating into artificial intelligence should be the road of intelligent manufacturing suitable for Chinese enterprises

Copyright © 2011 JIN SHI