The new transistor simulates the execution of a single neuron
The new transistor simulates the execution of a single neuron
According to the physicist organization network 20 reported that China and Singapore scientists in cooperation with the use of molybdenum disulfide to create a new type of neuron transistors". Each transistor simulates a single neuron in the brain that performs computational tasks and can be a basic component of various classes of neural hardware. Related papers are published in the latest issue of nanotechnology magazine. Transistors with the ability to perform weighted and threshold operations like neurons can be called neuron transistors". Weighted threshold operation refers to the signal of single neuron to receive to many other neurons and these signals were calculated, and then obtain the weighted gross threshold comparison, decide whether to activate some neural responses. The human brain has tens of billions of neurons, each of which performs many weighting operations and threshold comparisons per second to regulate human consciousness and behavior. In the new study, scientists at the University of Electronic Science and technology in Chengdu and Nanyang Technology University in Singapore, from a class called "new two-dimensional semiconductor materials of transition metal sulfides in the selected to replace the conventional transistor silicon molybdenum disulfide, made by functional simulation of single transistor element of god. The research team of neuron transistor neural functions were verified, the results show that the neurons in the synchronous transistor can control two logic gates, weighted arithmetic functions, and completed a similar calculation task by abacus, and has obvious advantage in speed. The study also found that neurons before other researchers create transistors, usually only less than 0.05 Hz frequency operation is performed, the frequency is much lower than the human brain neurons activated neural responses required 5 Hz this average value, and the access to the MoS2 neural transistor, to work in a frequency range of 15 Hz to 0.01 Hz, so it has superiority in the development of neural There is nothing comparable to this hardware. The research team hopes, then add more logic gates to neuron transistors, create a neuron model with biological function is more practical, then memristors these neurons to analog transistor integrated synapses, construct simulated brain function artificial neural system.
