We use deep learning to predict colours from nano-structured surfaces. Using data acquired in the Plasmonic Colours work, we trained Deep Neural Networks to predict the colours that result from different laser settings or different nanoparticle geometries. This allows us to accurately predict the outcome of an experiment or simulation without having to run the experiment or the time-consuming simulation, saving both time and money.
We’ve developed a new method for inverse design using deep learning. Given a colour, this method iteratively seeks to find what nanoparticle geometry or laser settings will give this colour. The method is straight forward to implement and effective.
This work has been published in Nature Scientific Reports