What is (Quantum) Machine Learning?

Introduction to Quantum Computing

Machine learning is a thing-labeler, essentially.
   -- Cassie Kozyrkov; Chief Decision Scientist at Google

Machine Learning is a completely different programming paradigm. Quantum Machine Learning is the use of quantum computing for the computation of machine learning algorithms.  Instead of giving explicit instructions to the computer, the program is achieved by using examples.  The machine learning algorithm finds patterns in data and turns those patterns into instructions. 

Machine learning can be computed using the three main ways of classification, regression, and segmentation.  In classification, the discrete label of a thing is predicted.  Regression finds a function that predicts the relationship between some input and the dependent output.  Segmentation partitions a population into groups with similar characteristics. 

Machine learning algorithms all have the three components of representation (rules, decision trees, support vector machines, etc. that represent the values of the knowledge), evaluation (a function to evaluate the algorithm performance: accuracy, prediction, cost, etc.), and optimization (search process to generate algorithm parameterizations).  

Machine learning can be divided into three subfields:

• supervised learning: given labeled photos of cats/dogs, identify new photos of cats/dogs
• unsupervised learning: given a collection of unlabeled data, learn some structure of the data
• reinforcement learning: given access to rewards based on our actions, aim to maximize the expected rewards