Physical neural network of metasurfaces trained without digital models
A Nature Communications article describes a physical neural network with four programmable metasurfaces that learns through electromagnetic interactions, without relying on digital models.
According to the summary recorded in Radar, a Nature Communications article presents a physical neural network made of four programmable metasurfaces. According to the source, the system learns through electromagnetic interactions to focus waves, recognize objects and perform localization. The central point is the training approach, which does not require digital models. The source does not give the exact date of the original article, performance metrics or implementation details, so check those in the journal itself.
To consult and verify, find the original article in Nature Communications by the title given and check authors, date, methods and results before citing it. This item is a general scientific finding with no task for a Rota Nacional user, so it is not a guide. If you use AI to study the topic, do not paste internal papers, unpublished research data or personal information about staff or participants; send only public excerpts.