Intel’s new brain inspired neuromorphic AI chip contains 8 million neurons, processes data 1K times faster

On Monday, Intel announced the Pohoiki Beach, a neuromorphic system comprising of 8 million neurons, multiple Nahuku boards and 64 Loihi research chips. The Intel team unveiled this new system at the DARPA Electronics Resurgence Initiative Summit held in Detroit.

Intel introduced Loihi in 2017, its first brain inspired neuromorphic research chip. Loihi applies the principles found in biological brains to computer architectures. It enables users to process information up to 1,000 times faster and 10,000 times more efficiently than CPUs for specialized applications like sparse coding, graph search and constraint-satisfaction problems.The Pohoiki Beach is now available for the broader research community and they can experiment with Loihi.

“We are impressed with the early results demonstrated as we scale Loihi to create more powerful neuromorphic systems. Pohoiki Beach will now be available to more than 60 ecosystem partners, who will use this specialized system to solve complex, compute-intensive problems,” says Rich Uhlig, managing director of Intel Labs.

According to Intel, Pohoiki Beach will enable researchers to efficiently scale novel neural inspired algorithms such as sparse coding, simultaneous localization and mapping (SLAM) and path planning. The Pohoiki Beach system is different in a way because it will demonstrate the benefits of a specialized architecture for emerging applications, including some of the computational problems hardest for the internet of things (IoT) and autonomous devices to support.

By using this type of specialized system, as opposed to general-purpose computing technologies, Intel expects to realize orders of magnitude gains in speed and efficiency for a range of real-world applications, from autonomous vehicles to smart homes to cybersecurity.

Pohoiki Beach will mark a major milestone in Intel’s neuromorphic research, as it will lay the foundation for Intel Labs to scale the architecture to 100 million neurons later this year.

Rich Uhlig says he, “predicts the company will produce a system capable of simulating 100 million neurons by the end of 2019. Researchers will then be able to apply it to a whole new set of applications, such as better control of robot arms.”

Ars Technica writes that Loihi, the underlying chip in Pohoiki Beach consists of 130,000 neuron analogs—hardware-wise, this is roughly equivalent to half of the neural capacity of a fruit fly. Pohoiki Beach scales that up to 8 million neurons—about the neural capacity of a zebrafish. But what perhaps is more interesting than the raw computational power of the new neural network is how well it scales.

“With the Loihi chip we’ve been able to demonstrate 109 times lower power consumption running a real-time deep learning benchmark compared to a GPU, and 5 times lower power consumption compared to specialized IoT inference hardware. Even better, as we scale the network up by 50 times, Loihi maintains real-time performance results and uses only 30 percent more power, whereas the IoT hardware uses 500 percent more power and is no longer real-time,” says Chris Eliasmith, co-CEO of Applied Brain Research and professor at the University of Waterloo

As per the IEEE Spectrum, Intel and its research partners are just beginning to test what massive neural systems like Pohoiki Beach can do, but so far the evidence points to even greater performance and efficiency, says Mike Davies, director of neuromorphic research at Intel.

“We’re quickly accumulating results and data that there are definite benefits… mostly in the domain of efficiency. Virtually every one that we benchmark…we find significant gains in this architecture,” he says.

Going from a single-Loihi to 64 of them is more of a software issue than a hardware one. “We designed scalability into the Loihi chip from the beginning,” says Davies. “The chip has a hierarchical routing interface…which allows us to scale to up to 16,000 chips. So 64 is just the next step.”

According to Davies, Loihi can run networks which are immune to catastrophic forgetting and can learn more like humans. He proved this with an evidence of research work done by the Thomas Cleland’s group at Cornell University, that Loihi can achieve one-shot learning. That is, learning a new feature after being exposed to it only once.

Loihi can also run feature-extraction algorithms immune to the kinds of adversarial attacks that can confuse image recognition systems. Traditional neural networks don’t really understand the features they’re extracting from an image in the way our brains do. “They can be fooled with simplistic attacks like changing individual pixels or adding a screen of noise that wouldn’t fool a human in any way,” Davies explains. But the sparse-coding algorithms Loihi can run work more like the human visual system and so wouldn’t fall for such shenanigans.

This news brings a lot of excitement amongst the community and they are awaiting to see a system that will contain 100 million neurons by the end of this year.

https://twitter.com/javiermendonca/status/1151131213576359937

https://twitter.com/DSakya/status/1150988779143880704