Can AI solve the mysteries of photonic nanostructures? - Advanced Science News

Researchers at Georgia Institute of Technology have demonstrated the use of artificial intelligence (AI) in obtaining valuable insights to the operation of photonic nanostructures, which manipulate light for applications such as signal processing, communications, and computing. The study was recently published in the journal Advanced Intelligent Systems. By proper selection of the geometrical features of these nanoelements, a large range of system-level functionalities (e.g., filtering, lensing, frequency conversion) can be achieved. While most reports on using AI techniques in the field of nanophotonics are focused on the design and optimization of nanostructures, such as finding the geometrical features of meta-atoms, the new approach seeks to use the "intelligence" aspects of AI to understand the physics of these nanostructures, for example, in assessing the feasibility of a response from a given nanostructure. This new approach is implemented in two steps: in the first step, the relation between input and output of the nanostructure is highly simplified by dimensionality reduction.