ML4IoT: A Framework to Orchestrate Machine Learning Workflows on Internet of Things Data

Abstract

Internet of Things (IoT) applications generate vast amounts of real-time data. Temporal analysis of these data series to discover behavioural patterns may lead to quali ed knowledge affecting a broad range of industries. Hence, the use of machine learning (ML) algorithms over IoT data has the potential to improve safety, economy, and performance in critical processes. However, creating ML work ows at scale is a challenging task that depends upon both production and specialized skills. Such tasks require investigation, understanding, selection, and implementation of speci c ML work ows, which often lead to bottlenecks, production issues, and code management complexity and even then may not have a nal desirable outcome. This paper proposes the Machine Learning Framework for IoT data (ML4IoT), which is designed to orchestrate ML work ows, particularly on large volumes of data series. The ML4IoT framework enables the implementation of several types of ML models, each one with a different work ow. These models can be easily con gured and used through a simple pipeline. ML4IoT has been designed to use container-based components to enable training and deployment of various ML models in parallel. The results obtained suggest that the proposed framework can manage real-world IoT heterogeneous data by providing elasticity, robustness, and performance.