Quantum neural networks achieving quantum algorithms

Abstract

This paper explores the possibility to construct quantum algorithms by means of neural networks endowed with quantum gates evolved to achieve prescribed goals. First tentatives are performed on the well known Deutsch and Deutsch-Jozsa problems. Results are promising as solutions are detected for different sizes and initializations of the problems using a standard evolutionary learning process. This approach is then used to design quantum operators by combining simple quantum operators belonging to a predefined set.