Antoine Desmet is the analytics program for surface equipment at Komatsu (then Joy Global). Antoine Desmet’s role is to deliver high machine reliability by detecting which components are about to fail, and predict their remaining useful life to help schedule timely maintenance.
From very technical beginnings, Antoine Desmet has moved into a lead role: developing an analytics strategy that delivers value to both their business and customers. From raw sensor data, all the way to maintenance insights; his team-mates and he is delivering the value-chain that he has architected.
Antoine Desmet oversees the development, execution and publication of analytics that detect patterns or anomalies in huge datasets, acquired from our mining plants. His “machine learning” algorithms first teach themselves how the mining machine works, and then use this knowledge to detect abnormal patterns. This “unsupervised” approach is simple, automated and mass-producible.
Antoine Desmet’s responsibilities also include Research and Development: anything from coding a custom Hidden Semi-Markov Model, to deep learning using Convolutional Neural Networks (CNNs) and LSTMs in TensorFlow. Antoine Desmet has supervised the Research and Development projects of 7 interns: it’s one of the most rewarding aspects of his work, especially when he gets to present our papers at international conferences. Antoine Desmet consider himself an expert in anomaly detection in time-series data, using machine learning. He has used Research for nearly a decade, and he is also adding Python to his arsenal. Along the way, he has acquired a lot of domain knowledge in industrial process analytics, reliability engineering, and mining in general.
Antoine Desmet caught the “data mining” bug during his PhD at Imperial College London, where he had to make sense of results hidden in millions of database records. There, Antoine Desmet built a Neural Network to perform real-time guidance and traffic optimisation. Antoine Desmet’s background is electronics engineering (major in software and networks). He has earned three degrees from three different countries.
Antoine Desmet will be presenting on Mining the miners’ data: The hunt for lurking maintenance issues within the Technology, Inovation & Operational Excellence conference at IMARC. Download the programme.