Since the power consumption of artificial intelligence is doubled by 2026, the international researcher team is inspired by the brain while dramatically reducing AI environmental footprints. We announced an ambitious roadmap of Ting Technology.
Comprehensive review, Released on January 22 at Nature A neurological computing that imitates the architecture of the brain by 23 major experts in academic and industries will revolutionize everything from smartphones to smart city, using only a small part of the energy of the conventional system. I have an overview of how to do it.
Gert Cauwenbergh, a prominent professor at the UC San Diego Bio Engineering Bureau and co -author of the dissertation, said:
The field seems to have reached an important moment. “We have a big opportunity to build a new architecture and open framework that can be developed in commercial purposes,” said DHIREESHA KUDITITITHIDI, a label on Robert F. McDarmott donation chairs and papers at the University of Texas University. Says. A corresponding author.
Potential applications range from scientific computing, artificial intelligence to expansion reality, wearable devices, smart agriculture, and urban infrastructure. Recent breakthroughs have already demonstrated technology promises. In 2022, the team led by CauWenberghs developed a chip that can execute a variety of AI applications using more energy than conventional systems, maintaining equivalent accuracy.
However, in order to achieve a widespread recruitment, researchers have claimed that neurological systems need to be more emulated with nerve connection, one of the main features of the brain. The human brain first forms a lot of connections before strategically eliminating most of them strategically. This is a process that optimizes both space efficiency and information retention.
“Extended scalability and excellent efficiency are derived from the large -scale parallelity and hierarchical structure of nerve expression,” says CauWenberghs, a dense local connection with a system in a core unit (such as brain gray -white). Explains how to combine sparse long -distance communication (similar to White). Project).
The impact may be considerable. In the Sandy Ego Super Computer Center, where the new computing architecture is regularly evaluated, researchers may be immeasurable. “This publication shows a very large possibility for the use of large -scale nerve computing for real applications,” he said, co -author of the center data -compatible science computing director. Amitava majumdar said.
Roadmap requires a wide range of collaboration between academia and industries for success, so that in addition to the development of user -friendly programming tools, a wider range of developers and researchers can access technology. I am emphasizing.
Instead of proposing a single solution, the author assumes a nerve form -designed spectrum adjusted to various applications. This flexible approach helps to promote hiring in various fields, from healthcare, robot engineering to environmental monitoring and autonomous systems.
It is already in progress. Last year, CauWenberghs and Kuditipudi are 4 million from the National Science Foundation to release the Neuromorphic Commons, the Neuromorphic Commons, Secure, open access to nervous computing hardware and tools We will provide and promote joint innovation in this rapidly evolving field.
If you find this work convenient, consider working with small, once, or monthly donation. With your contribution, we will enable you to continue to bring you the science and medical news that you can think of, accurate and accurate. Independent reporting requires time, effort, and resources, and your support can continue to explore important stories for you. At the same time, we can reach important discovery and development to the people who need them most.
