Background

Last updated Jun 29, 2022

• Type note
• Status needs revision
• Source book

Source:: 2021 Machine Learning with Quantum Computers

Quantum Machine Learning looks at the opportunities that the current development of quantum computers opens up in the context of intelligent data mining.

• On small scales, nature behaves radically different from what our intuition teaches us. We need another mathematical framework to describe its behavior. This mathematical framework is called Quantum Theory.
• A computer whose computations can only be described with the laws of Quantum Theory is called a Quantum Computer.
• The most famous language in which quantum algorithms are formulated is the Circuit Model.
  • It contains Qubit and Quantum gates that perform computations on those qubits.
• Error Correction is crucial for quantum computers in order to preserve quantum coherence throughout thousands of computational operations.
  • But it turns out to be much more difficult for quantum computers as compared to the classical ones.
• In the meantime, Noisy intermediate-scale quantum devices are being used.
  • These are the first prototypes of what may one day be a full quantum computer.
  • They have no Error Correction and therefore produce only approximate results of computations.
• A good Machine Learning method is a complex interplay of different parts, such as the data, the model and the optimisation algorithm.

• The term Quantum Machine Learning came into use only around $2013$.
• In $2014$, Peter Wittek published a monograph with the title Quantum Machine Learning - What quantum computing means to data mining.
• From $2014$ onwards, the interest in the field increased rapidly.
• Today, it has established itself as an active sub-discipline of Quantum Computing research.

• A typology, introduced by Aimeur, Brassard and Gambs ,¹ distinguishes four approaches of how to combine Quantum Computing and Machine Learning.
  • CC refers to classical data being processed classically.
    • In this context, it means machine learning based on methods borrowed from quantum information research.
    • Example: Application of tensor networks to neural network training, developed for quantum many-body system.
  • QC