Most popular NVIDIA explains how AI has an impact

NVIDIA explains how AI has an impact

David Hogan, senior director of NVIDIA enterprise, talked about how the company views the impact of the application of AI at this year's AI Expo

original title: NVIDIA explns how 'true option' of AI is making an impact

David Hogan, senior director of NVIDIA enterprise, talked about how the company views the impact of the application of artificial intelligence at this year's Expo

at the theme meeting, entitled "what is the real adoption of artificial intelligence", hogan provided a real example to illustrate how this technology was used and enabled by NVIDIA GPU. But first, he emphasized the momentum we see in the field of artificial intelligence

"many governments have announced their investment in AI and how they will position themselves," Hogan commented. "Countries around the world are beginning to invest in very large infrastructure."

the world's most powerful supercomputer is driven by NVIDIA's GPU. At present, the fastest ORNL summit uses an incredible 27648 GPUs, providing more than 144 petaflops. AI requires a lot of computing power, which makes NVIDIA in a good position to capitalize

"the computing demand of artificial intelligence is huge, which exceeds what anyone has seen in the standard enterprise environment before," Hogan said. "You can't train neural networks on a standard CPU cluster."

NVIDIA originally developed graphics cards for games. Hogan said that although this is still an important part of the company's business, the company began to turn to artificial intelligence as early as 2012

most of the speech is about autonomous vehicle, which is not surprising considering the demand and NVIDIA's expertise in this field. Hogan emphasized that you can't use CPU to train driverless cars at all, and provided comparisons in terms of cost, size and power consumption

autonomous vehicle manufacturers need to train Pb level data every day, repeat their models, and deploy them again in order to bring these vehicles to market

NVIDIA has a machine called dgx-2, whose computing speed can reach 200 billion times per second. "This is equivalent to 800 traditional servers in a box."

NVIDIA has 370 driverless cars. Hogan said that these cars cover most car brands in the world. Many of these companies are investing heavily, competing to launch at least "class II" driverless cars within the time frame of

"we have an automatic fleet," Hogan said. "We have no intention of competing with Uber, Daimler or BMW, but the best way we can help our customers achieve this goal is to try it ourselves."

"all the work done by our customers is done by ourselves, so we understand the challenges and how to do it."

the impact of the real world

Hogan pointed out that AI is a "cross organizational horizontal ability" and "the enabler of many things". It is undoubtedly a challenge to put forward some examples of industries that cannot be improved to a certain extent through artificial intelligence

after autonomous vehicle, NVIDIA expects the next large-scale artificial intelligence to appear in the medical field (of course, our dear readers already know this)

Hogan provides a natural example of the National Health Service (NHS) in the UK, which has a large number of patient data. Integrating these data and using artificial intelligence to interpret them can unlock valuable information. New material technology can be integrated with nanotechnology, biotechnology and information technology to improve medical care

AI can make medical imaging the same as or even better than some doctors. However, for most people, they are still two-dimensional images

Hogan showed how AI can transform two-dimensional images into three-dimensional models of organs that are easier to understand. In the following GIF diagram, we see that an X-ray photo of the heart is converted into a 3D model:

we also heard how artificial intelligence can help the field of genomics and help find treatments for human diseases. NVIDIA GPU is used in the miniature handheld device of Oxford nanopore, which can sequence DNA of plants and other objects in the field

in a blog post last year, NVIDIA explained how minit uses artificial intelligence to perform basic operations:

nanopore sequencing measures tiny ion flows through nanopores. When DNA passes through these pores, it can detect the change of signal. This captured signal produces raw data, which requires signal processing to determine the sequence of DNA bases - called "sequence". This is the so-called "softbag"

this analysis problem is very suitable for artificial intelligence, especially recursive neural networks. Compared with previous methods, RNNs can obtain higher accuracy in time series data, and Oxford nanopore sequencer is famous for this. "

Hogan pointed out that in many ways, e-commerce paved the way for artificial intelligence. Data collected for things such as advertising can help train neural networks. In addition, e-commerce companies have been committed to improving and optimizing their algorithms to attract customers, such as recommendations

"all these data, all the Facebook information we create, enable us to train networks," Hogan said

physical retailers are also improving through artificial intelligence. Hogan takes Wal Mart as an example. The company is using artificial intelligence to improve its demand forecasting and keep the supply chain running smoothly

in real time, Wal Mart can see where the potential supply challenges are and take actions to avoid or minimize them. The company can even see where the weather conditions may cause problems

Hogan said that this saved Wal Mart tens of billions of dollars. "This is just an example of how artificial intelligence has an impact today, not only on the profits of enterprises, but also on the overall performance of enterprises."

Hogan said that Accenture currently detects about 200million cyber threats every day. He pointed out that without AI, it is impossible to protect yourself from so many evolving threats

"this is impossible to solve. Look at it, give priority to it, and take any other way instead of applying AI," Hogan commented. "AI is based on patterns - different things - and when to take action when vanadium prices are low and when not to take action."

although we often hear about what AI will be used for one day, Hogan's speech provides an interesting perspective on how NVIDIA views the impact of AI today or in the near future

original text:

Introduction to the main functions of hydraulic pressure testing machine