XAI + AutoML – new shiny toys
During the talk I will present some new results in the area of eXplainable Artificial Intelligence and Automated ML.
In the first part I will overview methods for local model exploration for tabular data (SHAP, iBreakDown, aspect importance and modelStudio https://github.com/ModelOriented/modelStudio) with real world applications to credit scoring.
In the second part I will overview the Elo based interpretable measure of model performance and it’s applications to automated ML based on https://arxiv.org/abs/1908.09213.
Bio
I am interested in predictive modeling of large and complex data, data visualization and model interpretability. Currently my main research project is DrWhy.AI – tools and methods for exploration, explanation and debugging of predictive models. Other research activities are focused on applications, mainly high-throughput genetic profiling in oncology.
I work as Principal Data Scientist at Samsung SRPOL, associate professor at Warsaw University of Technology and assistant professor at University of Warsaw. Previously I was working in R&D departments of Netezza, IBM, iQor and Disney.