Backpropagation with Photonics

Abstract

This chapter presents an experiment that was not originally planned as part of my thesis. The project was set up when Michiel Hermans joined our team in 2015 with an idea of implementing the backpropagation training algorithm (more on that in Sect. 3.2) in hardware, using our opto-electronic reservoir computer (see Sect. 1.2.4) with one slight modification. After a couple of brainstormings, we decided that the idea would be easier to realise with a FPGA in the setup. I was thus set to assist Michiel Hermans with this experiment, that we successfully completed in March 2016. Michiel Hermans designed the setup and performed most ofthemeasurements, while I built the experiment and programmed the FPGA. The content of this chapter is based on our paper [1]. The paper itself, being a Letter, is quite succinct, but the SupplementaryMaterial covers most ofthe theoretical and experimental aspects of this work. The only part not covered in the paper—the FPGA design—is presented in this thesis in Sect. 3.4. 3.1