Event Details

Dr Thomas Schmid is a Lecturer for Computing at the newly established Lancaster Campus in Leipzig, Germany. His academic research focuses on developing machine learning tools and applied data science.

Calendar entry: 24 February, 12:30

Speaker: Thomas Schmid, University of Leipzig

Title: Constructivism as a Paradigm for Explainable Data Mining

Abstract: After decades of basic research, artificial intelligence (AI) still faces fundamental challenges. While human cognition is widely recognized as inspiration and role-model for AI systems, ambiguity and diversity in human cognition is a topic often paid little attention to in AI research. Implications of the modern philosophical and educational paradigm of constructivism, e.g., are rarely considered algorithmically. To this end, we suggest a hybrid AI approach that realigns the use of existing machine learning tools with respect to constructivist concepts. Considering classifiers as models with limited validity, this concept provides mechanisms for abstraction and differentiation of such models. From a data mining point of view, our framework represents a metadata-driven approach for extracting knowledge and organizing extracted knowledge by hierarchical and interpretable relationships.

Join: via Teams