Intel’s neuromorphic computing system learns like a human

Intel's new neuromorphic chip-powered system is a massive leap for artificial intelligence

Artificial intelligence and machine learning are in their infancy, and there is a lot of work ahead to perfect them. A few big hurdles to clear include processing time and power consumption. Intel Corp. has been hard at work developing its neuromorphic computing systems and is set to unveil its new Pohoiki Springs system that’ll solve both problems at once.

Advertisement - Article continues below

The Pohoiki Springs system is different from traditional computing systems in that its memory and computing systems are intertwined, while traditional systems separate them. This eliminates the shipping of data back and forth between the two areas, dramatically speeding up its processing time.

For example, Intel’s research team used just one neuromorphic research chip to train an AI system to detect a hazardous odor. This system learned to detect each odor after just a single training sample, while the traditional deep-learning method required 3,000 samples per odor.

How’s this possible? The neuromorphic computing systems learn much like an infant by permanently learning sensory-based objects – those they can see, smell, hear or touch – after only one interaction, according to Mike Davies, director of Intel’s Neuromorphic Computing Lab. What’s more, this system’s ability to learn instantaneously allows it to make predictions more accurately than a traditional deep-learning machine.

Advertisement
Advertisement - Article continues below

The Pohoiki Springs system boasts about 770 neuromorphic chips in a chassis that’s no larger than a standard server. This gives it the computational capacity of about 100 million neurons, which is about the same as a rat mole’s brain.

Advertisement - Article continues below

The byproduct of doing everything quicker and more efficiently is it requires less power to deliver the same results as a deep-learning machine. Power consumption has been a key impediment for large-scale AI development, as researchers at the University of Massachusetts found that developing just one AI model can have the same carbon footprint as five automobiles over their entire lifetime.

This advancement is coming at a \n ideal time, as research firm Gartner Inc. expects neuromorphic chips to be the predominant architecture for all AI deployments by 2025. There’s a good chance Intel will have a big hand in this move.

Featured Resources

Top 5 challenges of migrating applications to the cloud

Explore how VMware Cloud on AWS helps to address common cloud migration challenges

Download now

3 reasons why now is the time to rethink your network

Changing requirements call for new solutions

Download now

All-flash buyer’s guide

Tips for evaluating Solid-State Arrays

Download now

Enabling enterprise machine and deep learning with intelligent storage

The power of AI can only be realised through efficient and performant delivery of data

Download now
Advertisement

Recommended

Visit/technology/artificial-intelligence-ai/354766/mit-develops-ai-tech-to-edit-outdated-wikipedia
artificial intelligence (AI)

MIT develops AI tech to edit outdated Wikipedia articles

13 Feb 2020

Most Popular

Visit/security/cyber-crime/355171/fbi-warns-of-zoom-bombing-hackers-amidst-coronavirus-usage-spike
cyber crime

FBI warns of ‘Zoom-bombing’ hackers amid coronavirus usage spike

31 Mar 2020
Visit/security/data-breaches/355173/marriott-hit-by-data-breach-exposing-personal-data-of-52-million
data breaches

Marriott data breach exposes personal data of 5.2 million guests

31 Mar 2020
Visit/development/application-programming-interface-api/355192/apple-buys-dark-sky-weather-app-and-leaves
application programming interface (API)

Apple buys Dark Sky weather app and leaves Android users in the cold

1 Apr 2020
Visit/data-insights/data-management/355170/oracle-cloud-courses-are-free-during-coronavirus-lockdown
data management

Oracle cloud courses are free during coronavirus lockdown

31 Mar 2020