Quantum Operations for Natural Language Processing

点击翻译按钮获取中文摘要

Natural Language Processing (NLP) is a branch of Artificial Intelligence that aims to help computers understand human languages. Applications of NLP are commonly found in everyday life, e.g. in search engines such as Google and chatbots such as Alexa and Siri. Despite the success of these applications, NLP still faces limitations. These include the large amount of input data, and hence the computational time associated with certain tasks. Furthermore, current models of natural language used in NLP are not perfect: computers' output is usually very distinct from human responses, as one often witnesses in chatbot responses and Google's translations of sentences. Somewhat related to Quantum Machine Learning, Quantum Natural Language Processing (QNLP) is a recent field of research that aims at exploiting the advantages of quantum computers for NLP applications. Indeed, quantum computers are known to be able to outperform the best-known classical algorithms solving certain problems, including optimization problems. In addition, the motivation for QNLP goes further than the mere computational considerations. Indeed, some quantum-like models of cognition and natural language have surfaced, which represent a considerable argument for quantum artificial intelligence.In this project, we will focus on aspects of natural language which have a direct analogue with concepts of quantum theory, e.g. ambiguity and contextuality. These could then be used to come up with practical use of quantum computers for NLP. The project will be mainly theoretical; however, if theoretical results allow it, we will also aim at implementing QNLP algorithms on small- scale quantum computers.