Comparison between multi-class classifiers and deep learning with focus on industry 4.0

Abstract

Growing amounts of data will be one of consequences in Industry 4.0. This paper deals about mining frequent patterns and important factors in data. Classification is one of the most common assignments in data analytics. We used letter recognition data from the UCI repository as data set for our experiment. Data set contains more than 20000 instances of 26 classes. In our case, it represents multi-class classification. This idea can be transformed into industrial environment. Deep learning is a new area of machine learning research. We decided to use Deep learning from open source H2O machine learning framework and compare it with four multi-class classification algorithms available as services on Microsoft Azure. We are focusing this idea on Industrial systems, cloud architecture and data analytics, which will be fundamental pillars of Industry 4.0.