The recent explosion in the amount of available data and concomittant rise of machine learningMachine Learning (ML) (ML) has considerably broadened the range of available options to predict the properties of organic materials. The present chapter introduces the ML techniques currently used toward this goal. The emphasis is on methods to obtain numerical representations of the materials, called descriptors, features or vector embeddingsVector embedding in the MLMachine Learning (ML) context, rather than on the wide range of techniques applicable to regress properties of interest against those descriptors. Domain knowledge descriptors, supervised deep neural networks trained end-to-end from symbolic molecular representation to endpoints, and self-supervised and further-trainable representations for transfer learningTransfer Learning (TL) are subsequently discussed. Applications of these methods to predict properties of organic materials are illustrated through representative examples.