UPEC has published three papers about HyperCOG in the last year, highlighting the role that machine learning will play in achieving the final goals of the project.
The first of these papers, published in Sensors, propose a machine learning and data-driven methodology, based on data mining and clustering, for automatic identification and characterisation of the different ways unknown systems can behave. Applied to real Industry 4.0 data, this approach allowed the researchers to extract some typical, real behaviours of the plant, while assuming no previous knowledge about the data.
The second paper, published for the IEEE International Workshop on Intelligent Data Acquisition and Advanced Computing Systems: Technology and Applications, explores the possibilities of how to automate the initial data preparation and knowledge extraction tasks involved in machine learning, which at present requires experts to manually identify and characterise the studied systems – a long, exhaustive, and expensive task.
The final paper again discusses the benefits of machine learning algorithms, and presents some new avenues of reflection for automatic behaviour correctness identification through space partitioning, and density conceptualisation and computation.