This survey paper covers about three decades of research on pattern discovery. We consider some of the major breakthroughs about constraint-based data mining and several nice results about interesting pattern set discovery. We then focus on recent approaches for a cross-fertilization between pattern discovery and neural-based Machine Learning approaches. Specifically, neural methods have been developed for pattern mining and pattern set mining, enabling more efficient exploration of structural motifs in complex datasets. Additionally, hybrid approaches integrating Explainable AI (XAI) techniques have emerged, aiming to enhance the interpretability and transparency of machine learning models. We finally illustrate to what extent our colleague Arno Siebes has contributed to the considered topics.