Research: Exploiting quantum features for machine learning applications
The PhD research project fits directly into the research line of Quantum Information by exploring one of the most promising and active branches: quantum machine learning. The use of quantum mechanics to implement artificial intelligence problems is one of the great open problems of quantum computing. Given the possible practical applications, the discovery of a 'quantum advantage' in this branch would have huge implications.
Although this branch of research is only in its infancy, there are indications that the very structure of quantum states could lead to a significant acceleration of problems related to artificial intelligence. The states of a quantum system are associated with a vector (Hilbert) space, this is naturally completed with the notion of the norm of states and of the distance between two states. The calculation of the distance between two "items" is the basis of any classic classification protocol; thus, it is natural to ask whether intrinsic quantum mechanical distance could be used to speed up classification problems.
Another interesting example is the Quantum Variational Algorithm (QVA). In this case, you run some quantum code and update the quantum circuit parameter to arrive at the desired solution, for example, that by minimizing a certain cost function.
This has a clear analogy with the neural network used in computer science, but it also has a strong correlation with the variational procedure used in quantum mechanics to calculate the properties of atoms and molecules.
The central objective of the PhD project is, therefore, to use these properties of quantum mechanics to speed up some computational problems present in machine learning.