摘要
新闻编辑们从发明者提供的背景信息中获得了以下引文：“组织越来越多地实施Mac Hine Learning，以利用人工智能为各种应用提供实用价值。然而，从头开始开发这种实用的机器学习应用程序涉及许多耗时和无效的程序。”包括建立用于创建机器学习模型的环境，获取测试数据，培训和优化机器学习模型，发布优化的机器学习模型，以及与他人共享机器学习模型。例如，具有TensorFlow和Keras等库的重要知识的程序员必须投入大量时间，以便从头开始建立环境。一旦建立，优化机器学习模型需要人工和耗时的过程,这可能会导致结果不一致。此外,发布优化机器学习模型通常是一个人工过程。这些人工过程往往会带来不必要的模型风险,因为潜在的训练数据不一致、库版本差异以及其他人为错误等。团队成员之间共享机器学习模型通常是通过手动复制文件和库来实现的，这既耗时又容易出错。
Abstract
The following quote was obtained by the news editors from the background informa tion supplied by the inventors: “Organizations are increasingly implementing mac hine learning to take advantage of artificial intelligence to derive practical v alue for a variety of applications. Developing such practical machine learning a pplications from scratch, however, involves a number of time consuming and ineff icient procedures, including setting up environments for creating a machine lear ning model, acquiring testing data, training and optimizing the machine learning model, publishing the optimized machine learning model, and sharing the machine learning model with others. For example, a programmer with significant knowhow with libraries such as TensorFlow and Keras, must devote a significant amount of time in order to set up an environment from scratch. Once set up, optimizing th e machine learning model requires manual and time consuming processes which may cause inconsistent results. Further, publishing optimized machine learning model s is usually a manual process. These manual processes often introduce unnecessar y model risk because of potential training data inconsistency, library version d ifference, and other human errors etc. Furthermore, sharing of the machine learn ing model among team members is conventionally achieved by manually copying file s and libraries, which is both time consuming and error prone.
NETL
NSTL