Machine learning's broad use has been limited since the 1950s due to computational and data constraints. This work exploits expert knowledge to address these issues, introducing two sensor fusion methods and a multi-stage perception framework that combines traditional machine learning and deep learning. It utilizes knowledge graphs for simpler classifiers, creating explainable models that can be selectively retrained, thus optimizing resources. A technique for tailoring neural networks for industrial applications is presented, reducing data requirements.
«
Machine learning's broad use has been limited since the 1950s due to computational and data constraints. This work exploits expert knowledge to address these issues, introducing two sensor fusion methods and a multi-stage perception framework that combines traditional machine learning and deep learning. It utilizes knowledge graphs for simpler classifiers, creating explainable models that can be selectively retrained, thus optimizing resources. A technique for tailoring neural networks for indus...
»