Neuromorphic computing tries to make computers work like the human brain. It's about being smarter, smaller, and using less power. A review paper published in Nature by 23 researchers maps out how to make this happen. The researchers say we need different neuromorphic solutions for different uses, like artificial intelligence (AI), health care, and smart cities.
As AI grows, it uses more power. Neuromorphic chips could solve this by being more efficient. They could do more with less space and energy. In a press release issued by UC San Diego, Gert Cauwenberghs says this tech is crucial now because traditional AI systems are too power-hungry.
In 2022, a chip from Cauwenberghs' team showed how neuromorphic chips can be versatile and accurate with low power use. The NeuRRAM chip does AI work right in memory, saving energy.
Cauwenberghs explains that these systems need massive parallelism, like the brain's neurons, with local and global connections, to work efficiently.
“This publication shows tremendous potential toward the use of neuromorphic computing at scale for real-life applications," says Amitava Majumdar from the San Diego Supercomputer Center. This collaborative work "paves the path for bringing a neuromorphic resource for the national user community."
Toward commercial applications of neuromorphic computing
To scale up, neuromorphic systems should mimic the brain's sparsity. The brain grows many connections then prunes them for efficiency. Doing this in computers could make them more compact and energy-saving.
The researchers also suggest more collaboration and easier programming languages to grow the field. This would help people from different areas work together more easily.
“Neuromorphic computing is at a pivotal moment, reminiscent of the AlexNet-like moment for deep learning,” says Dhireesha Kudithipudi in a press release issued by UT San Antonio. “We are now at a point where there is a tremendous opportunity to build new architectures and open frameworks that can be deployed in commercial applications."
