Off the wire
• U.S. aircraft carrier strike group heads toward Korean Peninsula  • Feature: U.S. dignitaries say U.S., China must keep good relations to benefit world  • China Focus: Chinese companies come, hiring more locals in U.S.  • Strong working relationship between President Xi and Trump critical: Aust'n ex-politician  • Serbia beats Spain in Davis Cup quarter-finals  • Pakistani forces seize explosive-laden vehicle: army  • 1st LD Writethru: Bomb attack kills at least 9 pro-gov't militiamen in N. Afghanistan  • Roundup: Myanmar strives to boost foreign investment in new fiscal year  • Urgent: Bomb attack kills at least 9 pro-gov't militiamen in N. Afghanistan  • China's GDP likely to grow 6.8 pct in Q1: Goldman Sachs  
You are here: 　 Home

Chinese science academy invests in deep learning processor

Xinhua, April 9, 2017 Adjust font size:

The Chinese Academy of Sciences (CAS) has earmarked 10 million yuan (1.4 million U.S. dollars) for the research and development of an advanced AI processor.

The deep learning processor chip, the "Cambrian," is expected to be the world's first processor that simulates human nerve cells and synapses to conduct deep learning, according to a statement issued by CAS.

The program is named after the Cambrian Period, which marked a rapid diversification of lifeforms on earth. Scientists are expecting the processor to spearhead a new era in AI.

The statement said that the investment would be used in basic research areas to explore the structure and algorithm for the next generation of AI, which will lay the foundations for China's ambition in the global chip market.

The funds will also be used to promote the Cambrian.

The Cambrian research team is led by Chen Yunji and Chen Tianshi from the CAS Institute of Computing Technology.

Google's AI program AlphaGo needs huge power and large servers to operate, but the Cambrian aims to perform at the same level and use just one watt of power and be the size of a smartphone or a watch, according to Chen Yunji.

AlphaGo scored a 4-1 victory over Korean Go master Lee Se-dol last year. Endi