Perhaps the first reaction of most people is DeepMind, or Yandex of Russia. In fact, in France, there are many artificial intelligence enterprises that have been impressed by the romantic country in the past.
France is famous for Sophie Marceau and fashion week, sounds and technology can be completely detached, on the supply of artificial intelligent enterprise growth of the soil, the French are not lost in other European countries.
What is France, the data, the talent and the capital?
To say that the necessary conditions for large-scale development of AI enterprises are the following: first, we need massive data. Second, universities need to supply and supply as talents. Third, we need capital and government support.
In terms of data, Europe, which has a relatively small population and widespread mobile devices, is facing weakness in the face of Sino US and even India. But the advanced technology in some industries has also given France a new way.
Perhaps the most famous of France is the red wine, the film and the romantic feelings, but also in this country also has the geek and the modernized side. Grenoble, France, has the reputation of "European Silicon Valley". It has gathered a large number of semiconductor and microelectronics enterprises, and has hatched a series of related enterprises in IT, embedded electronics and other industrial chains. In Sophia, France, and the largest European Science Park established in 70s in the last century. These companies have given the French enough industrial data and engineering basis. In this way, although France can not gain the advantage in deep learning such data dependent algorithm, it can find a foothold in the industrial foundation and hardware capabilities.
In Colleges and universities, France has the Thackeray University of Paris, which combines the National Institute of science and research, the Institute of higher research and other scientific research institutions. France has also launched a "210 engineering college line" similar to China's 211 and 985, and many of these colleges have opened artificial intelligence related specialties. However, from the number of published papers and research results, there are still a lot of gaps compared to American colleges and universities.
Perhaps because France has a strict system of exam oriented education, in the college preparatory course, linear algebra, calculus such a partial theoretical assessment is very strict. But the present artificial intelligence is more dependent on probability and statistics, leading to a slight lag in the academic study of artificial intelligence in France.
But now the hot convolutional neural network, is actually by the French Le Kun Jahn (LeCun) is proposed. When he studied French in France, he put forward the concept of artificial neural network. Unfortunately, under the coincidence of chance, until recently, convolution neural network was utilized and valued. The music at the University of Toronto and the United States have Kunming veteran enterprises for many years, the French far distance. Or maybe he could bring a lot of deep - learning cows in France.
But in the government's support, France is stronger than many countries. Since 2010, France has begun to pilot innovative enterprises. In 2013, the French robot development plan was launched with the European Union's robot research and development program. In the summer of this year, France also introduced the French artificial intelligence strategy. Although it has not been able to introduce too much tax preferential policies, it is more important to promote the commercialization of research results in institutes and colleges and universities by increasing the way of public and private cooperation.
In short, in the background of developing AI, France's advantages lie in robotics and engineering. Although it has government support, it is in a weak position in capital and talent supply.
What is the French AI enterprise like? Maybe more than you think
In the development of the enterprise, it also embodies many characteristics of the development background of French artificial intelligence.
First, due to the influence of the government, many advanced research achievements of artificial intelligence in France are introduced in cooperation with the "national word". In the merger of Thackeray University in Paris, a large number of research institutions went to universities, forming a university research institution -- the benign talent industry chain of incubation project.
The National Academy of Sciences, France, has led the world's first study of artificial nanneurons, the first to recognize numbers from different people. This research represents the very high degree of stability of magnetic oscillators, and the high stability magnetic oscillator means that we can build simulated biological neurons, quickly process real-time large data, and ultimately create more energy-efficient micro chips.
Second, in the context of the EU's advocacy of robot development and the foundation of its engineering capability, France has many advantages in 3D printing, robots, UAVs and so on. Both the government and the capital are willing to pay for these projects.
As a French company has launched aerated UAVs, Bordeaux traffic police have begun to use unmanned aircraft to assist in their work. France gets the most financing AI business, and it studies the CARMAT of artificial heart. They are committed to making large numbers of artificial hearts by using 3D printing and so on, and use algorithms to match the monement between artificial heart and human body.
Third, because of the advantage of French as a French speaking country, a large number of NLP related projects have been acquired by companies such as Google and Facebook. In the NLP study, there is a huge gap between different languages. Different language, voice tagging, participle method and common slang are different, which is one of the reasons for Google's low market share in the Russian market. In the age of artificial intelligence for the success of the data theory, the difference between language is more obvious. To improve a language related NLP technology, the direct acquisition of a start-up is the most cost-effective.
Like the dialogue App platform, Wit.ai was bought by Facebook after 21 months. There is also a company called Snips, which aims to build personal voice assistant for users by searching the local data of mobile terminals, and now has invested tens of millions of dollars.
In addition to the acquisition and investment, some enterprises have begun to enter, directly reaping talents. Under the guidance of Le Kun, the head of AI research institute, Facebook set up artificial intelligence laboratory in France, in addition to attracting talents, it also established a cooperative relationship with the French National Institute of information and automation, and jointly launched research and development projects.
From graduation to cash, the French government's dedicated dragon service
If we want to sum up the development of AI in France, we can conclude from the following points: "exit mechanism is perfect", "engineering capability is strong" and "government dominant tendency is obvious".
In this way, the AI business in France seems to do well: under the influence of Le Kun, some deep learning conferences also came to France, and the strength of AI in academic field was gradually strengthened. After graduating from college, students enter the research institute directly, enjoying the cloud data sharing platform established by the state, and the bonus of commercialization of research results is regarded as the first barrel of gold. After going out of the Research Institute, although the entrepreneurial atmosphere is no more than that of the United States, it can recruit talents at lower cost. After the completion of the project, they can find European and American enterprises or governments to pay the bill.
But the reality behind such high - quality conditions is that there is little hope of the birth of an artificial intelligence giant in France.
The gap between China and the United States will also hold India, Israel and Europe as a whole. But even in Europe, French discourse power is not the strongest. To accelerate the commercialization of technology in the laboratory can be seen as a strategic policy against big businesses and capital acquisitions, leaving the ownership of talents, technology and data to France itself. Otherwise, in the future, France's technologies such as driverless and Internet of things, which are closely related to people's daily life, are all in the hands of other countries.
The development route of artificial intelligence in France is to abandon the positive battlefield with more remuneration and higher risk coefficient, and assume a small role in the development of AI, so that it can save something related to its own safety. It has to be said that the French government has been reacting quite quickly compared with many countries. Many small and medium-sized countries like France are in an awkward situation in the age of AI, but there is no way to win more initiatives for themselves.