Artificial intelligence has become a giant game, then why do they stand out?

The concept of artificial intelligence was introduced in the middle of the last century, but the technology has been at a low ebb since then. Until recently, the concept of artificial intelligence has been warming up, has become the latest technology hotspot, Garnter also predicted that artificial intelligence will become 2014 one of the 10 major technology trends.

Technology giants have also favored this area, and entered the siege. Google's deep learning system has been able to recognize the face of the cat, IBM's supercomputer Watson is not satisfied with the race to defeat the human player, Microsoft co-founder Paul Allen can let the robot pass the high school biology test, Russian billionaire Dmitry Itzkov even plan to The real-world Avatar program was implemented in 2045, and Facebook was also unwilling to set up an AI lab.

These giants stumble for many years to progress, ordinary people always have the illusion, artificial intelligence is a variety of high-end high cold advanced. In fact, the logic behind it can generally be simplified to such a process:

Collect massive data;

Information processing (the so-called feature design in jargon), which becomes useful and universal;

By means of the algorithm, the model, the output result, the result feedback to the real environment.

From the above logic process, we can see that in this wave of artificial intelligence, the role of data play more and more important, technology giants by virtue of data, computing advantages will occupy the best opportunity. Google researcher Ilya Sutskever said that if the data is sufficient, the computational power is strong enough, the artificial neural network is deep enough, even without the "pre training" preprocessing, will receive good results.

Given the increasing importance of data, what are the characteristics of future data?

Navia Bae founder Beau Cronin, has been concerned about the field of artificial intelligence, in a science and technology article put forward their own views. Beau Cronin that future data will show three features:

The carrier of the data will be very small (such as gene, the carrier is very small, but the information is huge);

Data must be compiled with precise and rigorous algorithms and models (e.g., genetic information);

The cost of data acquisition will be more and more high (personal information security more and more attention, the government is likely to legislate to regulate information acquisition).

The application fields of artificial intelligence mainly include deep learning, large-scale distributed data mining based on cloud computing, parallel computing chip of simulating human brain, such as the hot unmanned vehicle, intelligent robot also belongs to the category of artificial intelligence. In combination with the characteristics of the future data mentioned above, the barriers to entry into the depth of learning are getting higher, and large companies already have a first-run advantage and may well dominate, and later are not suited to do so. So where might the latter find an artificial intelligence opportunity?

Highly intelligent Unmanned vehicle navigation Scheme: Unmanned vehicle operation is actually based on prepared maps and algorithms, but there are many places there is no such prepared for the unmanned vehicle "learning" map. So provide a highly intelligent unmanned vehicle map scheme. For example, with the help of slam (simultaneous Localization and mapping, synchronous positioning and mapping), the robot moved from an unknown location in an unknown environment, and positioned itself according to position estimation and map during the movement, At the same time, the incremental map is built on the basis of its localization to realize autonomous positioning and navigation of the robot.

Organizational Behavior interpretation and prediction: Through data monitoring and analysis, the paper constructs the information decision model in the organization. For example, financial data and other public data can be used to explain a company's behavior and to anticipate the company's next actions and developments. The same applies to manufacturing and business, to understanding industry development, evaluating industry performance, and developing optimization measures.

In fact, not that these two areas, such as Google's big companies do not, but this is not their main business, not dominate, so the latter still have the opportunity.

(Responsible editor: Mengyishan)