FB designing its own chips to analyse live video

2018-05-27T23:27:58+05:00 -DM

CALIFORNIA-Facebook is designing its own chips to help analyse live video.

The hope is that the chips can make the process of filtering real-time footage more energy efficient and not so costly, according to Bloomberg.

Current methods require 'a huge amount of compute power' to monitor every video, said Yann LeCun, Facebook's chief artificial intelligence scientist. 'Let's imagine someone users Facebook Live to film their own suicide or murder,' LeCun said at the Viva Technology industry conference in Paris, according to Bloomberg.

'You'd like to be able to take down that kind of content as it happens'.

Facebook has been ramping up its AI efforts so it can better detect users who may show suicidal tendencies.

The software scans for certain phrases that could be clues, such as the questions 'Are you ok?' and 'Can I help?'

That's after reports surfaced that users were broadcasting violent acts, such as suicides and murders, on Facebook Live.

Creating their own chips would allow Facebook to spot this kind of content faster and more efficiently.

LeCun said the chips wouldn't be 'completely new for Facebook,' as it has ventured into hardware previously.

Firms like Intel, Samsung and Nvidia have long dominated the space of AI chips.

But with the rise of technologies like speech recognition, augmented reality and virtual reality, more and more companies are making the leap into hardware.

'Facebook has worked on hardware before: it makes its own server design, motherboards, its own communications chips for data centers,' LeCun said.

It had been rumored for several months that Facebook was working on its own chips.

Bloomberg reported last month that the firm was building chips that could be used in its Oculus Go headset or a pair of smart home speakers.

A job posting on Facebook's corporate website seeks to hire a manager to head an 'end-to-end SoC/ASIC firmware and driver development organization'.

SoCs integrate all components of a computer on a single circuit. They're commonly used in smartphones.

ASICs chips are processors that are used for a specific application, such as bitcoin mining.

The job posting states that the firm is looking for 'expertise to build custom solutions targeted at multiple verticals including artificial intelligence/machine learning'.

By producing its own chips, Facebook can also better integrate them in various hardware products.

A smart home speaker requires artificial intelligence in order to process requests or detect and recognize faces.

Last month, reports surfaced that Facebook was designing two Facebook-branded smart speakers, codenamed Fiona and Aloha, which would eventually be renamed to Portal.

Both would have touchscreens and advanced features like facial recognition.

 

View More News