The times they are a-changin’: in 2016, widescale commercial application of artificial intelligence was still a faraway high-tech dream. The ChinaBang Awards did not even have a specialized category for this technology.
2018 has seen China’s AI companies rising to the forefront and this year’s ChinaBang winners have proven that the country has plenty to offer to the world. Here are the five winners of the 7th ChinaBang Awards in the category of Best AI.
SenseTime is the most valuable artificial intelligence startup in the world. In April, the deep learning developer secured $600 million financing round led by Alibaba.
The company owes its success to its talent—an area in which China still lags behind compared to developed countries. The company was founded by one of China’s most prominent AI scientists Prof. Tang Xiao’ou from the Chinese University Hong Kong.
“As an AI company with a strong academic background, SenseTime has more than 800 researchers, including more than 150 Ph.D. students from the world’s top school. It is the largest group of Chinese scientists in the field of AI in Asia which gave SenseTime a foundation for fast development,” SenseTime Senior PR Manager Chris Gao told TechNode.
Favorable national policies to support AI development is another reason why the company has reached this level. But China also has the advantage of a multitude of application scenarios, says Gao. Many new industries have developed and acceptance of fresh ideas is quite strong. SenseTime now supplies over 400 companies and government agencies with their technology.
One field of AI has been particularly successful in China is image and face recognition. Face++, also known as Megvii, defeated 15 AI giants in computer vision competitions including Google, Facebook, and Microsoft. The company, which has users in more than 200 countries, is another award-winner at ChinaBang.
“Face recognition is a relatively neutral technology, so it doesn’t have application value in just one industry,” Face++ Operation Director Wei Wenyuan told TechNode. “We choose the most suitable one and at the same time the most abundant one in data and scenarios: Finance, security, retail, mobile phones, logistics, real estate and other industries.”
Currently, face recognition technology is most widely used in security and surveillance: 32 provinces and cities in China have integrated intelligent features in their public security system, said Wei. But Face++ is not stopping there: video recognition, IoT, and robotics are the next step. The company has recently bought robotics company Aresbots and is developing a robot for Foxconn, the company most famous for manufacturing the iPhone.
Developing hardware for AI is harder than one would think. DeePhi stands out in chips and hardware architecture.
“Looking at technology realization, the threshold for hardware technology is higher than that of software,” DeePhi Senior Brand Director Ji Yun told TechNode. “The accumulation of technology and knowledge needed by employees is more complex, and the cycle to product realization is also long.”
Thanks to the rapid development of computing power, algorithms have been evolving faster. However, if we are to have bigger breakthroughs in applications, we need to revolutionize the optimization of hardware, says Ji.
During the second half of 2018, the company will launch its self-developed deep learning SoC chip called Ting (听涛). The company has received investments from US semiconductor product developer Xilinx, Alibaba’s financial arm Ant Financial, and from Samsung, one of the world’s largest chip maker.
Ping An is not a name one would connect with AI at first look; it is one of China’s largest insurance companies. But it turns out insurance plays well with AI. One example is car accidents: Ping An Technology uses image recognition to assess the damage to the car.
Ping An is also looking into other applications including medicine where image recognition for X-rays is used for diagnosis, customer support where voice recognition is used to assess the customer’s mood, and even in music. Ping An’s AI music won first place award the International AI Music Composition Competition hosted by Switzerland’s Federal Polytechnic School in Lausanne (EPFL) in January.
The final ChinBang awards winner is certainly a unique entrant to this list—it really gets to your brain. Westwell started with a splash in 2016 when it presented “Westwell Brain,” the first brain simulation to have 10 billion neurons with hardware. The company has developed the DeepSouth neural processor, a chip that simulates human brain neurons. DeepSouth is an answer to IBM’s own experiment inspired by the brain, the TrueNorth neuromorphic chip.
Westwell has branched out to other areas of medicine such as gene sequencing, health-focused wearables, and medical equipment. The company has recently moved to heavy machinery. Westwell Lab is developing products in industrial robotics, unmanned equipment for container terminals and ports, as well as autonomous vehicles.