Dr. Roy Assaf joined IBM Research Zurich in March 2018 where he currently works with the AI Automation group. His research is focused on machine learning and deep learning methods for predictive analytics, computer vision and on studying and improving explainability of deep learning models.
He is currently working on three projects; His main project is on bridge defect detection. He is developing a pipeline for detecting multiple defects using instance segmentation based on the Detectron2 computer vision library. he is also Developing a continuous learning semi-supervised approach for improving model accuracy and reducing the need for manual re-annotation of instances.
In his second project he is developing a deep learning pipeline for explainable anomaly detection of IBM storage devices. Here he developed a parallel pipeline which can process data from over 10000 storage systems efficiently. An essential aspect for applying deep learning for large-scale systems. His research aims at developing an explainability method that extracts impactful KPIs and ultimately enables support engineers to achieve quick root cause analysis.
In his third project (ROMEO https://www.romeoproject.eu/), he leads the effort where IBM and other industrial partners develop diagnostic and prognostic models for offshore wind turbines. His research involves the development of a seq-2-seq stochastic temporal convolution neural network (STCN) for improving the forecasting capabilities of existing state-of-the-art methods.
Roy holds a Ph.D. in Robotics, and throughout his education he has been awarded a Distinction Master of Science, and the prestigious European Marie Curie Research Fellowship for pursuing a Ph.D. degree. He is also the recipient of the Best Paper Award at the 2017 IEEE Prognostics and Health Management conference.