Imitating the way that humans gain knowledge, deep learning is a subset of artificial intelligence (AI) that goes beyond machine learning. Deep learning is based on artificial neural networks using representation learning, which is a set of techniques where a system can automatically discover the representations needed to detect features or classifications from raw data. Machine learning algorithms are usually categorized as supervised or unsupervised. Supervised learning utilizes a dataset that contains labeled observations or classes, whereas unsupervised learning explores the dataset and draws inferences to describe the hidden structure of unlabeled data. You can also have semi-supervised learning, where some observations of the dataset are labeled but most are not, so a mixture of supervised and unsupervised methods will be required.