Machine learning (ML) components are being added to more and more critical and impactful software systems, but the software development process of real-world production systems from prototyped ML models remains challenging with additional complexity and interdisciplinary collaboration challenges. This poses difficulties in using traditional software lifecycle models such as waterfall, spiral or agile model when building ML-enabled systems. By interviewing with practitioners from multiple companies, we investigated the application of using systems engineering process in ML-enabled systems. We developed a set of propositions and proposed V4ML process model for building products with ML components. We found that V4ML process model requires more efforts on documentation, system decomposition and V&V, but it addressed the interdisciplinary collaboration challenges and additional complexity introduced by ML components.

翻译：机器学习（ML）组件正被越来越多地集成到关键且影响重大的软件系统中，但从原型化ML模型到实际生产系统的软件开发过程仍面临挑战，尤其是其额外的复杂性和跨学科协作难题。这使得在构建ML赋能系统时，传统软件生命周期模型（如瀑布模型、螺旋模型或敏捷模型）的应用变得困难。通过与来自多家企业的从业者进行访谈，我们研究了系统工程流程在ML赋能系统中的应用。我们提出了一系列命题，并设计了V4ML流程模型用于构建包含ML组件的产品。研究发现，V4ML流程模型需要投入更多精力于文档编制、系统分解及验证与确认（V&V），但它有效解决了跨学科协作难题以及ML组件带来的额外复杂性。