About the project
In this project, you will develop and evaluate novel quantum machine learning approaches to solve large scale inverse problems using near term quantum computing systems.
By formulating inverse problems in a physics informed learning framework, efficient encoding of the data will be achieved, whilst at the same time allowing efficient hybrid model training. This framework also naturally allows for the inclusion of regularisation constraints.
Quantum physical principles provide an exciting new basis for the design of the next generation of computers. Based on the 4 basic postulates of quantum physics, these quantum computers utilise simple mathematical principles that allow us to define quantum states, their evolution, measurement, and integration to develop novel computational rules that allow the development of a wide range of novel algorithms. Due to the inherent nature of quantum parallelism, many of these approaches have been shown to efficiently solve several challenging computational problems.
Quantum computation has thus found a wide range of applications in machine learning. These advances have led to a re-evaluation of many traditional algorithms that run on classical computational hardware, with many novel Quantum Inspired algorithms leading to significant computational advantages even in classical settings.
In this PhD project, you will develop and evaluate novel quantum machine learning approaches to solve large scale inverse problems using near term quantum computing systems. By formulating inverse problems in a physics informed learning framework, efficient encoding of the data will be achieved, whilst at the same time allowing efficient hybrid model training. This framework also naturally allows for the inclusion of regularisation constraints.
This is a field where there is significant scope that allows you to follow your interests to pursuit different directions, whether these are theoretical, by looking at theoretical algorithm performance and convergence properties, or whether these are more practical, by applying these ideas to realistic tomographic data-sets from the fields of acoustic or X-ray tomographic imaging.
