Semiconductor company Intel introduces Loihi 2, a second generation neuromorphic research chip, and Lava, an open source software framework for developing neuro-inspired applications.
Corresponding Intel, Neuromorphic Computing, which uses insights from neuroscience to design chips that function more like the biological brain, seeks improvements in energy efficiency, computational speed and learning efficiency in a range of edge applications: from vision, speech, and gesture recognition to search retrieval, robotics and limited optimization problems.
The Loihi 2 is based on three years of first generation research and leverages advances in process technology and asynchronous design methods.
Advances in Loihi 2 enable the architecture to support new classes of neuro-inspired algorithms and applications while providing up to 10 times faster processing, up to 15 times higher resource density with up to one million neurons per chip, and improved energy efficiency offer, Intel claims.
Take advantage of Intel Technology development group, Loihi 2 was made with a pre-production version of the Intel 4 process, highlighting the health and advancement of Intel 4 technologies. This made it possible to develop Loihi 2 quickly.
According to Intel, the Lava software framework meets the need for a common software framework in the neuromorphic research community. As an open, modular, and extensible framework, Lava will allow researchers and application developers to build on each other’s advances and converge on a common set of tools, methods, and libraries.
Lava runs on heterogeneous architectures with conventional and neuromorphic processors and enables cross-platform execution and interoperability with a variety of frameworks for artificial intelligence, neuromorphic and robotics. Developers can start developing neuromorphic applications without access to specialized neuromorphic hardware and contribute to the Lava code base, including porting it to run on other platforms.
“Investigators at Los Alamos National Laboratory used Loihi’s neuromorphic platform to investigate the tradeoffs between quantum and neuromorphic computing and to implement learning processes on the chip, ”explains the research fellow at Los Alamos National Laboratory Dr. Gerd J. Customer.
“This research has shown some exciting equivalences between spiking neural networks and quantum annealing approaches to solving tough optimization problems. We have also shown that the backpropagation algorithm, a fundamental building block for neural network training that was previously thought to be impossible to implement on neuromorphic architectures, can be efficiently implemented on Loihi. Our team is excited to continue this research with the second generation Loihi 2 chip, ”explains Kunde.
Loihi 2 and Lava provide researchers with tools to develop and characterize new neuro-inspired applications for real-time processing, problem solving, adaptation, and learning. Notable highlights include:
Faster and more general optimization: Loihi 2’s programmability supports difficult optimization problems. This includes real-time optimization, planning and decision-making advantages for data center systems.
New Approaches to Continuous and Associative Learning: Loihi 2 improves support for advanced learning methods, including variations of backpropagation, the workhorse of deep learning. This expands the scope of adaptation and data-efficient learning algorithms that can be supported by energy-saving form factors operated in online settings.
Novel neural networks that can be trained through deep learning: Fully programmable neuron models and generalized spike messaging in Loihi 2 offer new neural network models that can be trained in deep learning.
Early evaluations indicate that with Loihi 2, compared to standard deep networks running on the original Loihi, over 60 times fewer ops per inference are reduced without loss of accuracy.
Integration with real robot systems, conventional processors and novel sensors: Loihi 2 addresses a practical limitation of Loihi by integrating faster, more flexible and more standard input / output interfaces.
Loihi 2 chips will support Ethernet interfaces, glueless integration with a wider range of event-based vision sensors, and larger mesh networks from Loihi 2 chips.
“Next-generation neuromorphic architecture will be critical to Accenture Labs’ research on brain-inspired computer vision algorithms for intelligent edge computing that could power future extended reality headsets or intelligent mobile robots,” says Accenture Labs scientific director and managing director Edy Liongosari.
“The new chip offers features that make it more efficient for hyper-dimensional computing and more advanced on-chip learning, while the Lava API gives developers a simpler and more streamlined interface for building neuromorphic systems,” he adds.
Intel says commercializing neuromorphs from laboratory research is a three-pronged endeavor.
It requires continuous iterative improvement of neuromorphic hardware in response to the results of algorithmic and application research; Developing a common cross-platform software framework so developers can compare, integrate and improve the best algorithmic ideas from different groups; and intense collaboration between industry, academia, and governments to build a rich, productive neuromorphic ecosystem for exploring commercial use cases that offer short-term business value.
Intel is currently offering two Loihi 2-based neuromorphic systems to dedicated INRC members through the Neuromorphic Research Cloud: Oheo Gulch, a single-chip early evaluation system, and Kapoho Point, an eight-chip system that will be available soon.
The Lava Software Framework is available for free download. to disposal GitHub. A presentation and tutorials on Loihi 2 and Lava will be featured at the upcoming Intel Innovation Event in October.
INTRODUCING ITWIRE TV
iTWire TV brings unique value to the tech sector by providing a range of video interviews, news, views and reviews, and also gives vendors the opportunity to promote your business and marketing messages.
We’ll work with you to develop the message and conduct the interview or product review in a safe and collaborative manner. In contrast to other tech YouTube channels, we create a story around your message and publish it on the ITWire homepage by linking to your message.
In addition, your interview post message can be displayed in up to 7 different post ads on our website iTWire.com to drive traffic and readers to your video content and downloads. This can be a significant lead generation opportunity for your company.
We also offer 3 videos in one recording / session if you wish so that you have a range of videos to promote to your clients. Your sales team can add your emails to sales collateral and to the footer of their sales and marketing emails.
See the latest in tech news, views, interviews, reviews, product promos and events. Plus funny videos from our readers and customers.