Fig. 2: Four different variants of machine learning (ML) algorithms.
From: Designing and understanding light-harvesting devices with machine learning
a In supervised learning, ML models can be used to directly predict properties of interest such as absorption spectra from molecular structures. b Unsupervised learning methods, such as clustering can be used to identify the most relevant information in a presented dataset. c Active learning approaches enable a ML model to query information during the training process. d Generative models can simultaneously predict molecular structures and properties of interest that go beyond prespecified training sets.
