Strongly Entangling Neural Network: Quantum-Classical Hybrid Model for Quantum Natural Language Processing

One of the most used techniques to improve a Machine Learning model is to gather more data. An interesting field in Machine Learning is Sequence Modelling, having Natural Language Processing as the peak of the field. The capabilities of Quantum Computing have been growing recently entering the novel field of Quantum Machine Learning. In this paper, we propose a Quantum Natural Language Processing classification model named Strongly Entangling Neural Network. This model leverages the quantum advantage to imitate part of the behavior of a Recurrent Neural Network to process text data into the circuit and perform the classification task. This is accomplished by representing our data in a quantum circuit that relies heavily on the entanglement property of qubits. The results of our model have very favorable metrics, particularly obtaining a 97.70% of accuracy.