Quantum Artificial Intelligence

Coordinators: Emmanuel Rousseau and Didier Felbacq

We are exploring the potential of quantum technologies to solve problems that are inaccessible to classical machine learning algorithms. In particular, we study the prospects offered by a model of recurrent neural networks, known as quantum reservoir computing.

This approach leverages the complex dynamics of a quantum system for information processing and learning, without requiring explicit training of all the internal components of the neural network. We use this theoretical framework to assess the performance of different physical qubits, such as superconducting qubits, luminescent defects in silicon carbide, and Rydberg atoms, in terms of memory capacity, robustness, and energy efficiency.

In a long term perspective, this work aims to develop new paradigms of quantum artificial intelligence, capable of harnessing the intrinsic properties of quantum mechanics -superposition, entanglement, and decoherence- to design more powerful and adaptive learning architectures.

This research field is based on the project « quantum reservoir for efficient signal processing » : https://www.qrc-4-esp.eu/