For years, research into machine learning algorithms was more focused on supervised learning techniques. As supervised learning techniques can be directly used for inference, their benefits in terms of time, cost, and accuracy are relatively easily measurable. The power of unsupervised machine learning algorithms has been recognized more recently. As unsupervised learning is not guided, it is less dependent on assumptions and can potentially converge the solution in any dimension. Although it is more difficult to control the scope and processing requirements of unsupervised learning algorithms, they have more potential to unearth the hidden patterns. Researchers are also working to combine unsupervised machine learning techniques with supervised learning techniques in order to design new powerful algorithms.