A Software Architecting for Quantum Machine Learning Platform in Noisy Intermediate-Scale Quantum Era

Abstract In the current Noisy Intermediate-Scale Quantum (NISQ) era, despite quantum computing is challenged by scale limitations and noise, it still holds immense potential to accelerate machine learning for specific problems. However, we have found that existing machine learning platforms do not adequately take into account the properties of quantum computing, resulting in a lack of an integrated quantum machine learning environment. Therefore, this article proposes a software architecture specifically designed for quantum machine learning platform, which aims to provide development tools that offer full life-cycle support for quantum machine learning. We first analyze the software requirements of quantum machine learning platform by using user story method, and then construct the platform architecture blueprint by using multi-view software architecture method. To validate the effectiveness of the architecture we proposed, we used this architecture as a specification to implement quantum reinforcement learning (QRL) and construct a quantum convolutional neural network (QNN). The results indicate that our architecture can effectively support the fusion of quantum computing and machine learning. We believe that this software architecture will propel the development of quantum machine learning in the NISQ era, bringing new opportunities to quantum computing.