A Deep-Learning-Based Proposal to Aid Users in Quantum Computing Programming


New languages like Open QASM and SDKs like QISKit open new horizons for the research and development in the new paradigm of quantum computing. Despite that, they present an evident learning curve that could be hard for regular developers and newcomers in the field of quantum computing. On the other hand, currently there are many ways to build intelligent systems that can learn from humans and processes to build a knowledge corpus and provide a different kind of help to humans in tasks like aiding in decision making processes, recommending multimedia resources, building conversational agents, etc. In this paper we describe a work-in-progress project developed by the IBM Q team that implements an intelligent system based on a deep learning approach that learns how people code using the Open QASM language to later offer help and guidance to the coders by recommending different code sequences, logical steps or even small pieces of code. During the paper, we describe our current approach and first results. They include the use of seq2seq neural networks that effectively learn quantum-code sequences, and which will be tested in real context in the near future to improve the user experience in IBM Q Experience products.

In International Conference on Learning and Collaboration Technologies. HCI International 2018. Learning and Collaboration Technologies. Learning and Teaching (pp. 421-430). Springer, Cham