Timothy Lanfear, NVIDIA’s European solution architecture and engineering team, gave an account of what Artificial Intelligence means for hardware processing and how it will be pivotal going forward to all industries, at the IET Cambridge Local Network webinar evening lecture.
He started the lecture by explaining how from the starting point of computer graphics for single computer use, we are all now in a position where we have multiple digital devices that we expect to do multiple different things. Our digital requirements and experiences have changed, all the way from, gaming, health care, robotics, smart cities, automation and IOT that all are ready for hardware enhancement with the use of A.I.
He spoke about how beyond hardware, and by moving further up the stack can we fully take advantage of the A.I ecosystem with advanced chip technology, as all the software development interfaces and toolkits that are also required to be able to make best use of the systems are required to be constructed within accelerated chip environments.
He spoke about how across these components A.I Training and A.I Inference is important to have a successful A.I system. From how we extract information from data sets using appropriate rules, from which, deep learning networks are trained to understand intensions within data sets, using algorithms from which neural networks can be established to perform trained tasks.
We are seeing increased A.I within our consumer and business products now, such as;
He spoke about how beyond hardware, and by moving further up the stack can we fully take advantage of the A.I ecosystem with advanced chip technology, as all the software development interfaces and toolkits that are also required to be able to make best use of the systems are required to be constructed within accelerated chip environments.
He spoke about how across these components A.I Training and A.I Inference is important to have a successful A.I system. From how we extract information from data sets using appropriate rules, from which, deep learning networks are trained to understand intensions within data sets, using algorithms from which neural networks can be established to perform trained tasks.
We are seeing increased A.I within our consumer and business products now, such as;
- Recommender Systems - OnDemand Video Recommendations (You know the one.. Want to watch the next episode now?)
- Smart Cities - Traffic light monitoring
- Health Care - Radiologists are in short demand, so A.I can help.
- Robotics - Automated Factories
- Autonomous Vehicles - Driverless cars
- Retail Assistance - Chat or instore virtual services.
Typically, the A.I systems learns with millions of parameters to give response in fractions of a second - research shows that a human must be able to do multiple queries per second for an A.I system to be 'satisfactory'.
He touched on how the Cambridge-1 AI supercomputer is a concrete example of A.I capability. This system, which will be available for U.K. healthcare researchers to work on pressing problems, is located at Kao Data in England, a data centre using 100 percent renewable energy.
He touched on how the Cambridge-1 AI supercomputer is a concrete example of A.I capability. This system, which will be available for U.K. healthcare researchers to work on pressing problems, is located at Kao Data in England, a data centre using 100 percent renewable energy.