New turbo chip inspired by AI neurons

NeuRRAM is the name of a new chip inspired by neurons that promises turbine-charged artificial intelligence: small and versatile, it consumes less power than conventional chips and performs precise computations directly in memory, without having to rely on an external cloud, thus offering a wide range of applications and better data protection sensitive. It could in the future be used in many devices, from smartwatches to virtual reality headsets, from sensors for Industry 4.0 to roving vehicles for space exploration. The study indicates this published In Nature by an international research team led by the University of California, San Diego.

Data processing in AI is very complex and energy-intensive: its hardware applications typically require data to be moved to the cloud, where it is analyzed and processed by AI to produce results that are then sent back to the device.

“It’s like going from eight hours from home to work for just two hours,” researcher Weir Wan explains. This happens because battery-powered devices have little power to devote to data processing. By reducing power consumption, the NeuRAM chip could pave the way for smarter, more reliable, and more accurate devices like traditional digital chips. NeuRRAM is very versatile and can support different types of neural networks: that is why it can be used for many applications, such as image reconstruction and speech recognition.