In 1957, American psychologist Frank Rosenblatt suggested an element of a neural computer network as perceptron . Perceptron - a three-level switch that can be trained to respond in predictable ways to certain events. Perceptron was not something unknown. In the form of complex elements perceptrons were used to build neural networks. However, the first time a potential singleton perceptron created just now.
Under the leadership of Vyacheslav Alexandrovich Demin - Director for Science Center Kurchatov NBIKS - a group of scientists from a number of Russian institutes and universities, as well as staff from the University of Parma in Italy, he published a scientific work devoted to the invention of perceptron based memristor. The work is called "Hardware elementary perceptron based on polyaniline memristive devices". Based on the conductive polymer polyaniline created a sample of organic nonvolatile memory element with the ability to self-study. According to scientists, the first time it became possible to talk about multi-layer and multi-level neural networks, high-density recording.
Prototype perceptron has large dimensions. It remains to be reduced to the production scale . Scientists say the plans to produce elemental base on perceptrons within the 10-nm process technology. It is obvious that it is a long time, although the 10-nm semiconductor promise to appear before the end of this year.
Related Products :
|