Presenter: Rastislav Fáber (PhD student at Slovak University of Technology in Bratislava)

 

In recent years, the integration of data analytics and machine learning (ML) algorithms into industrial processes has shown to be a promising approach in increasing production efficiency. By utilizing real-time data from industrial sensors and applying ML algorithms to classify the data, the production process can be more extensively monitored, providing a cheap validation guideline that would not have been possible otherwise. However, the performance of ML classifiers must be assessed based on the complexity of implementation to determine the most suitable model for decision-making. In our case, accurate categorization of measurements is critical to identify slight deviations from real values that may impact the quality of the final product. The most important factor in selecting a suitable classifier model is its ability to handle the outlier detection. Furthermore, it is important to note that the complexity of the data plays a crucial role in the performance of ML algorithms. By accurately categorizing real-time data from industrial sensors, it is possible to minimize human intervention in process control.