# Quantum Machine Learning

A very emerging new interdisciplinary research area in the field of Quantum Machine learning, a branch or intersection of Machine learning and Quantum physics. It is most commonly used for quantum-enhanced-machine learning. Quantum machine learning basically increases the capabilities of machine learning to compute immense data on machine learning algorithms by making way for opportunities to conduct various analyses on quantum states and systems. It involves both quantum processing and classical. A difficult computational** **Subroutine is mainly outsourced to a quantum device. The quantum devices help these routines to be executed faster as they can be complex in nature.

The term “quantum machine learning” is also associated with classical machine learning methods that are applied to the data generated from quantum experiments (i.e. *machine learning of quantum systems*), like for example: learning quantum phase transitions^{ }or for creating new quantum experiments. Quantum machine learning is also a very extended branch of research that explores methodological and structural similarities between learning systems and physical systems, mainly neural networks. For example, some of the mathematical and numerical techniques from quantum physics are also applicable to classical deep learning and vice versa. The main advantage of it is that it improves classical machine learning.

The algorithms of the given classical data are needed to be encoded by one and set into a quantum computer so that it can be accessible for quantum information processing. Then in the next step routines of quantum information processing applied and then the result of the quantum computation is read out by measuring the quantum system. Most of them have been implemented on a small scale or special-purpose quantum devices as most proposals of quantum machine learning algorithms are mostly purely theoretical and require a full-scale universal quantum computer to be tested. Some of its implications are:

- Linear algebra simulation with quantum amplitudes: Amplitude encoding is the main idea of what a number of quantum algorithms are based on. This is in order to associate the amplitudes of the quantum state with the outputs and inputs of computations.
- Quantum machine learning algorithms based on Grover Search: this approach uses amplitude amplification methods which are based on Grover’s search algorithm. Unstructured search problems with a quadratic speedup are solved with this as compared to classical algorithms. These quantum routines can be employed for learning algorithms that translate into an unstructured search task, as can be done, for instance, in the case of the k-medians and the k-nearest neighbor’s algorithms. Another application is a quadratic speedup in the training of perception.
- Quantum-enhanced reinforcement learning: here, a quantum agent is made to interact with a classical environment and occasionally receive rewards for its actions, which allows the agent to adapt the behavior thus, they get used to or learn what actions to do to earn rewards. The Implementations of these kinds of protocols have been proposed in superconducting circuits
^{ }and in systems of trapped ions. - Quantum annealing: it is mainly an optimization technique that is used to determine the local minima and maxima of a function over a given set of candidate functions. This is a basic method used to discretize a function that has many local minima or maxima, in order to determine the observables of the function.