Data Modalities and Representation Learning

Real-life data comes in many forms. Some of the well-studied data types include tabular data, texts, time series, and images. The field of unimodal machine learning focuses on the design and development of algorithms capable of operating on a single data type; an example is tree ensembles used on tabular data that represent house prices. This chapter serves as an overview of different data types commonly considered and some common approaches that are, at this point, widely used in academia and industry for solving a plethora of real-life problems.