Machine Learning for Industrial Internet of Things

Abstract

Machine learning is now a popular practise in a variety of fields and has permeated our daily lives. These approaches have been used in various fields, and their use is constantly growing. These methods are particularly important for supporting Industry 4.0 and IoT situations. Many of the algorithmic findings, however, cannot be comprehended or justified in terms of how or why a particular choice was taken. Few studies have been generated regarding the end-user perspective, despite the fact that a number of strategies and approaches have evolved in recent years as a result of the advancement of machine learning research. Therefore, the main barrier to the adoption of these applications is the lack of interpretability in this technology. Machine learning has a broad field called anomaly detection, which has a lot of applications in the sphere of industry. In reality, it is crucial for many different things, including quality control and preventative measures. The advantage of this strategy is that it may be applied without the need for labelled data, but it seems strange not to have labelled data in this kind of framework where the data is frequently "dirty." Obviously, the interpretability issue that the entire family faces also affects this final application.