AI will ‘fuel’ future IOT devices says Nvidia co-founder

Image credit: E&T

Artificial intelligence (AI) created from ‘deep learning’ initiatives will power the next wave of Internet-of-things (IOT) devices according to the co-founder of graphics company Nvidia.

Speaking at the company’s GTC 2016 conference in Amsterdam, Jen-Hsun Huang explained how deep learning, a process through which machines learn based on a set of algorithms rather than pre-coded behaviours, will bring unprecedented intelligence to everyday objects.

“Your vaccum cleaner is already relatively intelligent but it has the ability to be much more intelligent,” he said.

“Your toaster, your coffee maker, your house, the cameras that watch the outside of your house; these devices are going to be infused with AI so that they are much, much more intelligent.”

“I believe the AI era will connect tens of billions of devices to the internet. Only our imagination limits us.”

“This is the era of the intelligent device and deep learning is the fuel for IOT.”

Huang said that the ability for computers to perceive and learn without human guidance will shape the technology sector in the coming years. At the event, Nvidia announced it was releasing its Tesla P100 GPU accelerators that are designed to boost deep learning applications in data centres.

The chips are designed with energy efficiency and speed in mind and are based on Nvidia’s latest ‘Pascal’ architecture. It was demonstrated how ‘neural networks’, which link stacks of deep learning GPUs in order to create a system that works in a similar way to the human brain, could understand the content of pictures and even artistically reinterpret individual frames of a video (see below).

The system is capable of aping famous modern artists such as Picasso and Van Gogh in a similar fashion to Google’s DeepDream software.

Technology research firm Tractica recently released its AI Market Forecast report which predicts that AI will expand rapidly in the coming years.

“Tractica forecasts that annual revenue for enterprise applications of AI will increase from $358m (£275m) worldwide in 2016 to $31.2bn in 2025,” it said. “In total, the top 10 use cases will account for approximately 40 per cent of the overall AI software market revenue in 2025, representing $14bn in value."

Tractica's top 10 list of AI uses cases includes:                      

  1. Algorithmic trading strategy performance improvement
  2. Static image recognition, classification and tagging
  3. Efficient, scalable processing of patient data
  4. Predictive maintenance
  5. Content distribution on social media
  6. Text query of images
  7. Automated geophysical feature detection
  8. Object identification, detection, classification, tracking from geospatial images
  9. Object detection and classification - avoidance, navigation
  10. Contract analysis

Recent articles

Info Message

Our sites use cookies to support some functionality, and to collect anonymous user data.

Learn more about IET cookies and how to control them