6/05/2018

arm AI

EVP of arm

Humanizing AI

1 trillion devices in 2035 (from 11.4 billion now)

3+ billion smartphone

8.4 billion IoT devices

 

Global computing power. 2.16 trillion gigaflops now to 100 trillion gigaflops in 2035

 

Data and AI revolution

 

Concerns

75% of people concern that auto cars may be hacked

85% of people worry that AI may make personal data lost

57% of people worry that AI machine intelligent than human

60% of people worry that criminal use of AI technology

 

5G may help AI as the higher speed and lower latency

Millions of tiny moments 

Respiro smart module for healthcare industry

Ashma's inhaler with sensors: The devices will measure and upload the inhale so as to detect the effectiveness of inhaler. It helps a child with Ashma to be adjusted and recove soon.

Intelligence at the edge (5G edge computing)

 

The smartphone is the most popular devices.

Digital assistant, augmented Reality, …

ElliQ

Nauto for safer driving

Reindeer cam

Kissfly drone…

 

ML moving to the edge, why?

Reliability

Power

Cost

Security

Reliability

Bandwidth

 

Arm ML computing platform (Machine learning, object detection)

CPU, GPU, ML/OD processor, DSP/FPGA/accelerators.

Data center

Auto driving

Image enhancement

Voice and image recognition

沒有留言: