Research smart chip generation for phone learning

After Qualcomm, scientists at Purdue University in Indiana are also researching a generation of intelligent chip processors (AI chips) for phones that are capable of learning the brains of animals.

We have known that the development of AI chips has been developed by many technology companies for years, but the problem is that their size is large and is often used for supercomputers, workstations, not Targeted to mobile devices such as smartphones and tablets, smart chip design for mobile will open a new future for these devices, stronger, smarter and "understand" users. .

Eugenio Culurciello, a professor at Purdue University's team, said that developing an immediate learning chip would allow mobile devices to understand the content of video and images.

At the Neurobiology Information Processing System Conference in Nevada, the Purdue University team has introduced a compact co-processor that can be attached to a regular phone to enable it to run parts. Soft learning, this software is capable of recognizing the face or a scene of the road.

Their co-processor has been tested with an FPGA chip , which can be programmed to test on a new hardware-based phone line without much development cost.

At present, the team's research is just as simple, but Eugenio Culurciello believes that they can improve the processing power of the chip 10 times in the future, and he said the immediate goal of the group. is developing the platform and then selling the technology to interested companies such as Apple, Qualcomm, Samsung to equip their products.