By its very nature, genomics produces large, high-dimensional datasets that are well suited to analysis by machine learning approaches. Here, we explain some key aspects of machine learning that make it useful for genome annotation, with illustrative examples from ENCODE.