AI chipsets, solving current and future industry problems

#News(IoTStack) [ via IoTForIndiaGroup ]

Semiconductors power the electronically pervasive modern world. Artificial Intelligence (AI) is destined to support a vast number of technologies ranging from self-driving vehicles to social media analytics. The combination of AI with semiconductors represents a logical next step for silicon providers to embed AI into virtually anything that requires a processor says the author Gerry Christensen, founder and CEO of Mind Commerce.

AI chipsets also support the industry move towards heterogeneous computing refers to systems that use more than one kind of processor or cores. This approach involves integration of dissimilar chips in a co-processing environment to establish performance gains and save energy. In the case of AI chips, processors can often solve problems that traditional processors cannot.

Leading manufacturers are developing solutions to solve industry problems. For example, ARM is building AI into their Machine Learning (ML) Processor to speed-up critical functions for industry such as facial recognition for public safety and homeland security.

This same chipset can enhance other industry needs such as machine translation. This approach is part of System on a Chip (SoC) deployments that incorporate different processors types such as Graphical Processing Units (GPU), which optimise image identification calculations.

Some companies are focused on solving device specific issues such as Qualcomm and Huawei, which are both racing to deliver the first AI chipsets for smartphones. The companies recognise that network developments such as Mobile Edge Computing (MEC) and 5G will drive increasingly more data usage, much of which will be for latency-sensitive services such as Virtual Reality (VR). Handset manufacturers do not want to be the weak link in data processing.

Read More..