With the Loihi 2 second-generation research chip and open source Lava software framework released last year, Intel Research is leading the way in neuromimetic computing.
As part of Intel's goal to commercialize neuromimetic technology, Intel Research is providing developers with new tools to take the development process to the next level. The 8-chip Loihi 2 development board, Kapoho Point, for example, can be stacked to meet the needs of large-scale workloads and enables direct interconnection with low-latency event-based vision sensors.
Neuromorphic computer is a method of computer system design in which elements of a computer system are imitated systems in the human brain as well as nervous system. The term describes the style of both software and hardware computer components.
Neuromorphic engineers draw from several self-controls-- consisting of computer science, biology, math, electronic design as well as physics-- to create artificial neural systems inspired by biological structures.
There are two overarching goals of neuromorphic computer (in some cases called neuromorphic engineering). The initial is to create a tool that can learn, preserve info and even make logical deductions the method a human mind can-- a cognition machine. The 2nd goal is to get new info-- and perhaps show a logical theory-- concerning how the human brain works.
Intel Labs' second-generation neuromorphic research study chip, codenamed Loihi 2, and also Lava, an open-source software structure, will drive innovation as well as fostering of neuromorphic computer remedies.
As much as 10x faster handling capability1
Up to 60x more inter-chip bandwidth2
Approximately 1 million nerve cells with 15x higher source density3
3D Scalable with native Ethernet support
A brand-new, open-source software structure called Lava
Fully programmable neuron versions with graded spikes
Improved learning and also adaptation abilities
Kapoho Point, a Loihi 2-based development board, is a compact system ideal for a wide range of small form factor devices and applications from drones to satellites and smart cars. Kapoho Point can run AI models with up to 1 billion parameters and also solve optimization problems covering up to 8 million variables. It increases speed by more than 10 times and reduces energy consumption by a factor of 1000 compared to advanced solvers running on CPUs. In addition, Kapoho Point can be scaled by stacking multiple development boards to solve larger scale problems.
The U.S. Air Force Research Laboratory (AFRL) is the first member of the research community to enable Kapoho Point and is using it for internal research involving impulsive neural network-based learning and problems that require real-time optimization. Because AFRL's missions are conducted in the air and in space, this leaves the mobile platform with very limited space, weight and power budget," said Dr. Ching Wu, senior scientist for data processing and development. For the need to run AI algorithms in this environment, neuromimetic computing technology provides an excellent computational solution."
Updates to the open source, modular and scalable Lava software framework include a number of improvements for the Loihi 2 feature set, such as programmable neurons, hierarchical events and continuous learning.
Intel Research has updated the open source Lava framework to support programmable neurons, integer impulse neurons, convolutional networks, and continuous learning.
The Lava software framework is available for free download on GitHub.
The Intel Neuromimetic Research Community (INRC) has initiated eight Intel-supported university projects, including George Mason University, Queensland University of Technology, Graz University of Technology, University of Zurich, Brown University, Pennsylvania State University, University of Waterloo, and University of Göttingen.
Research projects include adaptive robot localization, wireless bionic sensing pulse decoding for brain-computer interfaces, neuromimetic Bayesian optimization, auditory feature detection, and novel brain-like architectures and algorithms.
Since its inception in 2018, the Intel Neuromimetic Research Community has grown to more than 180 members, including university labs, government agencies, and leading global companies such as Accenture, Lenovo, Logitech, and Mercedes-Benz.
Next, Intel Research will continue to provide developers with new tools to make it easier for them to develop applications that solve real-world problems and continue to support community research.