MOSCOW, April 25 Scientists from the Russian company Smart Engines have found a way to speed up the work of neural networks by 40% — they proposed a new working model to replace the existing 8-bit one; as the company’s general director Vladimir Arlazarov explained, this will reduce equipment costs and expand the pool of tasks performed by artificial intelligence.
«Deep neural networks are constantly becoming more complex, containing hundreds of millions or more coefficients, which require more computing power. This limits the use of central processors in artificial intelligence systems. Smart Engines researchers have solved this problem by proposing a qualitative improvement to the 8-bit model — 4.6- bit networks work 40% faster than the 8-bit model, but are almost as good in quality due to more efficient use of the features of the central processors of mobile devices,” the company said. 
«Fast and highly efficient AI is needed everywhere and by everyone today. Every person wants ChatGPT on their mobile phone. And 4.6-bit models are an important step on this path. They allow, on the one hand, to reduce the cost of equipment for already existing solutions. On the other hand, to solve a completely new class of computer vision problems on current equipment, where previously there were not enough computing resources,” Arlazarov explained.
Today, working with neural networks is possible on specialized video cards, but not all computers are equipped with them. But every user device — be it a computer or a smartphone — has a central processor, and for it the use of 8-bit neural networks is a global standard. As Smart Engines said, 4.6-bit neural networks are “lighter” and are easier to use in central processors on different devices.
The company already uses 4.6-bit neural networks in its developments, in particular to solve applied computer vision problems for searching and recognizing objects — they are usually performed on devices with low computing capabilities. In addition, the development can expand the class of tasks performed by the on-board computers of unmanned vehicles.
“As a result of the sanctions, mobile applications of leading banks were removed from stores. Then it was possible to create web applications that retained all the usual functions, including payments using QR codes. This was largely achieved thanks to 4.6-bit networks. .. Although this looks like a question of performance, it is actually a question of security. To do this, you need your AI to be with you, your data to be with you, that is, on your smartphone, and not on a huge server. to implement functionality on a small mobile phone and for it to work, special tricks are needed,” added Arlazarov.
Свежие комментарии