Computer vision
Computer vision refers to the computer's ability to recognize objects, scenes and activities from images. Computer vision technology uses a sequence composed of image processing operations and other technologies to decompose image analysis tasks into manageable small tasks. For example, some techniques can detect the edge and texture of an object from an image, and classification techniques can be used to determine whether the recognized features can represent a class known to the system.
Computer vision has a wide range of applications, including: medical image analysis is used to improve disease prediction, diagnosis and treatment; Face recognition is used by Facebook to automatically identify people in photos; It is used to identify suspects in the field of security and monitoring; In shopping, consumers can now take photos of products with their smartphones to get more purchase options.
As a related discipline, machine vision generally refers to the visual application in the field of industrial automation. In these applications, computers identify production parts and other objects in a highly restricted factory environment, so the goal is simpler than computer vision that seeks to operate in an unrestricted environment. Computer vision is an ongoing research, while machine vision is a "solved problem", which is a subject of system engineering rather than a research level. Because of the expanding application scope, some start-ups in the field of computer vision have attracted hundreds of millions of dollars of venture capital since 20 1 1.
machine learning
Machine learning refers to the ability of a computer system to improve its performance only by relying on data without following explicit program instructions. Its core is that machine learning automatically discovers patterns from data, and once patterns are discovered, they can be used for prediction. For example, if the machine learning system is given a database of credit card transaction information, such as transaction time, merchant, location, price and whether the transaction is legal or not, the system will learn a pattern that can be used to predict credit card fraud. The more transaction data is processed, the more accurate the forecast will be.
Machine learning has a wide range of applications, and it may improve almost all the performance for those activities that generate massive data. In addition to fraud screening, these activities include sales forecasting, inventory management, oil and gas exploration and public health. Machine learning technology also plays an important role in other cognitive technology fields, such as computer vision, which can improve its ability to recognize objects by constantly training and improving visual models in massive images.
Nowadays, machine learning has become one of the hottest research fields in cognitive technology, and attracted nearly 1 1 ~ 20 14 dollars of venture capital. Google also spent $400 million in 20 14 to acquire Deepmind, a company that studies machine learning technology.
Invested $400 million to acquire Deepmind, a company that studies machine learning technology.
natural language processing
Natural language processing refers to the computer's text processing ability similar to that of human beings. For example, extracting the meaning from the text, or even interpreting the meaning independently from those readable texts with natural style and correct grammar. A natural language processing system does not understand the way human beings process texts, but it can skillfully process texts with very complicated and mature means. For example, automatically identify all the people and places mentioned in a document; Determine the core issues of the document; In a pile of human-readable contracts, various terms and conditions are extracted and made into tables. Traditional text processing software can't accomplish these tasks, and can only operate on simple text matching and patterns.
Like computer vision technology, natural language processing combines various technologies that help to achieve the goal. Establish a language model to predict the probability distribution of language expression, for example, the maximum possibility of a given string of characters or words expressing specific semantics. The selected features can be combined with some elements in the text to identify paragraphs of the text. By recognizing these elements, you can distinguish certain types of text from other words, such as spam and ordinary mail. The classification method driven by machine learning will become the standard of screening, which is used to determine whether an email belongs to spam.
Because context is so important to understand the difference between "timeflies" and "fruitflies", the practical application fields of natural language processing technology are relatively narrow, including analyzing customers' feedback on a specific product and service, automatically discovering some meanings in civil litigation or government investigation, and automatically writing formulaic short articles on corporate income and sports, and so on.
robot
Cognitive technologies such as machine vision and automatic planning are integrated into extremely small but high-performance sensors, brakes and cleverly designed hardware, thus giving birth to a new generation of robots, which have the ability to cooperate with human beings and can flexibly handle different tasks in various unknown environments. For example, drones, "cobots" that can share work for humans in the workshop, etc.
speech recognition
Speech recognition mainly focuses on the technology of automatically and accurately transcribing human speech. This technology must face some problems similar to natural language processing, and there are some difficulties in dealing with different accents, background noise and distinguishing homophones ("buy" and "by" sound the same). At the same time, you need to keep up with the normal speech speed. Speech recognition systems use some of the same technologies as natural language processing systems, supplemented by other technologies, such as acoustic models that describe sounds and their probability of appearing in specific sequences and languages. The main applications of voice recognition include medical dictation, voice writing, computer system voice control, telephone customer service and so on. For example, Domino_Pizza recently launched a mobile phone application that allows users to place orders by voice.
The industrialization of the above five technologies is the key to the industrialization of artificial intelligence. Artificial intelligence will be a trillion-dollar market, even 10 trillion-dollar market, which will bring us some brand-new and huge sub-industries, such as robots, smart sensors and wearable devices. The most worth looking forward to is the robot sub-industry.
There are many ways to divide robot applications, which can be roughly divided into the following categories from the application level. The first category is industrial robots. Companies like Foxconn make good use of it, because labor costs are getting higher and higher, and employment risks are getting higher and higher, and robots can solve these problems. The second type is the monitoring robot, which can be used as a nurse for patients, the elderly or children at home and in hospitals to help them do some complicated things. In fact, the demand for monitoring robots in China is even more urgent, because the demographic dividend in China is declining and the aging is on the rise. These two contradictions can be solved by robots. Therefore, the demand in this field accounts for a large proportion in the civil market. The third kind is the exploration robot, which is used for mining or exploration, and greatly avoids the danger that people have to experience. In addition, there are military robots used for fighting.
Online media Business Insider predicts that robots will replace humans in many positions: telemarketers, proofreaders, hand tailors, mathematicians, insurance underwriters, watch maintenance personnel, freight forwarders, taxpayers, image processors, bank account holders, librarians, typists and so on. Because their price competitiveness is amazing. The research of McKinsey Global Institute shows that when the manufacturing wage in China increases by 10% ~ 20% every year, the global robot price will decrease by 10% every year, and the cost of the cheapest low-order robot is only half of the average annual wage of Americans. Gu Neng, an international research institution, predicts that robots will lead to a new round of unemployment in the world in 2020.
At the same time, the development of artificial intelligence technology will also make many old industries look brand-new, the most typical of which is the automobile industry. The automobile industry has existed for hundreds of years, and the changes during this period are also very great, but people are always driving. In recent years, with the strong investment of companies such as Google, machines or some automation systems are expected to replace people to drive cars, thus forming a new industry with huge market capacity, namely the driverless car industry. The scale of this industry will also be trillions or even 10 trillion. Moreover, this industry will be superimposed and integrated with the new energy industry to form a composite industry of "vehicle networking+energy networking+Internet plus electric vehicles"-in the future, we will use plug-in vehicles and hydrogen fuel vehicles as power devices to make new energy vehicles a part of the power grid and become suppliers of new energy, just as some houses with solar power generation systems are solar suppliers now.
There is no doubt that intelligent technology will penetrate into almost all old industries like the Internet. Huatai Securities mentioned nine industries in a research report on artificial intelligence industry: life service O2O, medical care, retail, financial industry, digital marketing, agriculture, industry, commerce and online education. In fact, many old industries will be reborn, such as military, media, home, medical and health industries, life sciences, energy, public sector and even virtual industries affected by the development of VR/AR technology. (The content comes from the robot family)
Hope to adopt, thank you.