This book presents a comprehensive range of topics in deep learning for polymer discovery, from fundamental concepts to advanced methodologies. These topics are crucial as they address critical challenges in polymer science and engineering. With a growing demand for new materials with specific properties, traditional experimental methods for polymer discovery are becoming increasingly time-consuming and costly. Deep learning offers a promising solution by enabling rapid screening of potential polymers and accelerating the design process. The authors begin with essential knowledge on polymer data representations and neural network architectures, then progress to deep learning frameworks for property prediction and inverse polymer design. The book then explores both sequence-based and graph-based approaches, covering various neural network types including LSTMs, GRUs, GCNs, and GINs. Advanced topics include interpretable graph deep learning with environment-based augmentation, semi-supervised techniques for addressing label imbalance, and data-centric transfer learning using diffusion models. The book aims to solve key problems in polymer discovery, including accurate property prediction, efficient design of polymers with desired characteristics, model interpretability, handling imbalanced and limited labeled data, and leveraging unlabeled data to improve prediction accuracy.