Brain-Inspired Nanowire Network Learns “On the Fly” for Efficient Machine Learning
Researchers at the University of Sydney and the University of California, Los Angeles, have developed a physical neural network that learns and remembers dynamically. The neural network, comprised of tiny nanowires, mimics the neural networks found in the brain. It was able to perform tasks by responding to changes in electronic resistance at points where the wires intersect, in a similar way to synapses in the brain. By recognizing and recalling sequences of electrical pulses the network was able to execute tasks such as image recognition, using dynamic data accessed online, avoiding heavy memory and energy usage.
Lead author Ruomin Zhu joined Technology Networks in an exclusive interview to discuss the findings of their study and how it may impact the future of artificial intelligence.
