The magic of AI lies in application

The magic of AI lies in application
Needless to say, artificial intelligence or AI has become a hot topic these days, thanks to advances in hardware, more cutting-edge algorithms and huge amounts of available data. However, the real magic of AI lies in application – when applied properly, AI can benefit a wide range of verticals from transportation to healthcare.
 
That was the key argument raised by speakers during an AI forum alongside Computex, held June 5 to 9 in Taipei.
 

AI enablers

 
Several driving forces have propelled AI to the next level. The first is hardware, which has become more advanced than ever. Chips like GPUs are now powerful enough to run complex deep learning algorithms that enable AI. “When we talk about training, the CPU is no longer sufficient because they are built for general purposes. In AI applications we've got to have something special just for training and inference purposes. Therefore we need accelerators. These could be GPUs or FPGA,” said Tau Leng, SVP of Technology and Marketing at Super Micro.
 
Data, meanwhile, is another driver with which systems are trained to recognize a wide range of objects. “Data is very important. We’ve had deep neural networks for the past 20 to 30 years, but now we have data. The more data we have, the higher accuracy you have,” said Winston Hsu, Professor of Department of Computer Science and Information Engineering at National Taiwan University.
 

Applications

 
However, it is the application of AI that should matter to end users, and the speakers shared how it can be applied in different vertical markets. In transportation, for example, AI can teach autonomous vehicles to recognize different objects. “You’ve got lots of sensors, such as LIDAR, RADAR, among others. You have all these sensors that collect data, which is then used to train the system to recognize what's a car, what's a license plate, what's a red light. That takes a lot of computing,” said Deepu Talla, VP and GM of Autonomous Machines at NVIDIA, which now has the NVIDIA DRIVE AI platform that allows automakers to build and deploy self-driving cars, trucks and shuttles that are functionally safe and can be certified to international safety standards.
 

Manufacturing

 
Another area where AI can come in handy is manufacturing. One example is machine vision where the system is trained with large amounts of images and videos on defects. This way it will be able to tell when a defect comes up during inspection, making the process more efficient for factory workers who can concentrate on other revenue-generating operations. “In our experience, we are able to help clients improve productivity. Production value per employee rose 16.9 percent while production value rose by 16.3 percent,” said Allan Yang, CTO of Advantech.
 

Healthcare

 
Finally, healthcare is a major vertical market that can benefit from AI. Arm, for example, has processors that can be embedded to medical devices such as asthma inhalers, which can run machine learning algorithms that study patient behavior or frequency of usage. When a patient deviates from an established set of patterns an alert will be sent. This can make the life of patients, their family and doctors much easier.
 
“You can really start to see this is going to change people's lives. Whether it's self-driving cars that are safer, whether it's drug companies that provide medicine, whether it’s patients that live healthier … the implications for us are significant,” said Rene Haas, President of Intellectual Property Products Group at Arm.
 
Daniel Tse, Product Manager for Medical Imaging at Google, meanwhile discussed their AI solutions that are trained to detect factors leading to diabetes-induced blindness and breast cancer. This way treatment can start early, increasing the patient’s survival rate.


Product Adopted:
Other
Share to:
Comments ( 0 )