Jan van Lunteren received the M.Sc. degree in Electrical Engineering, the MTD (Master of Technological Design) degree in Information and Communication Technology, and the Ph.D. degree in Electrical Engineering in 1992, 1994, and 1998, respectively, all from the Technical University of Eindhoven, The Netherlands. He has been with the IBM Zurich Research Laboratory, Switzerland, since 1994, performing research on machine learning, near-memory processing, pattern scanning for intrusion detection, high-performance programmable accelerator engines and network processors.
As a member of the Data & AI systems group, Jan is currently responsible for designing and implementing accelerated CPU-based machine learning inference functions targeted at IBM Z mainframes, IBM POWER systems, and x86-based computer systems. He has developed and invented new concepts for scoring decision-tree-based models, such as gradient boosting and random forests, that substantially outperform state-of-the-art solutions by competition. By supporting the import of pre-trained models from other machine learning frameworks, such as XGBoost and lightGBM, the scoring of those models can also be accelerated using Jan’s inference engine.
Jan’s work is integrated within the Snap Machine Learning (SnapML) library, which is available through IBM Watson ML community edition.