In the spring of this year, the company’s engineers introduced the Google Brain artificial intelligence AutoML able to create their own unique AI without human assistance. Not so long ago it became known that AutoML for the first time created a computer vision system NASNet far greater than all the analogs made by man. This system is based on AI can be a great help in the development of, for example, Autonomous cars, and robotics, allowing to bring the vision of robots to a whole new level.
AutoML develops according to the so-called system of reinforcement learning. In fact, it is a management network which develops a completely new neural networks for specialized tasks. In this case, the main purpose AutoML was the establishment of a system for maximally accurate recognition of objects in video in real time. AI alone has trained a new neural network, by tracking errors and making adjustments to its work. The learning process was repeated many thousands of times until, until the system became operational. Moreover, it surpassed all existing today similar to the neural network created and trained people.
According to the official statement by Google, the recognition accuracy NASNet equal or 82.7%. This is 1.2% better than previous record set in September by experts from Oxford and Momenta of the company. The neural network also turned out to be 4% more efficient analogues with 43.1% average precision. A simplified version of the NASNet, adapted to the mobile platform surpasses similar neural networks by more than 3%. This system may in the future be used to create Autonomous cars, because for them computer vision is incredibly important. Meanwhile AutoML continues to create new neural networks, and who knows what heights he will be able to achieve in the near future.
AI Google created its own AI that surpasses all counterparts
Sergey Grey