Identifying Topology of Photonic Lattices with Machine and Deep Learning

Topological photonics paves the way towards efficient optical processing of information due to the particular robustness of topological states of light against various forms of disorder [1]. This robustness can be characterized by the quantized topological invariants. Thus, extracting topological invariants constitutes an important task in diagnostics of experimental samples from both the fundamental and applied perspectives. Here, we propose an alternative to traditional methods of probing the topological invariant (such as band tomography) by using instead machine learning techniques [2].