Title: Interpretable Machine Learning

Abstract: 

(1) Interpretability and Machine Learning

Fundamentals of machine learning. Classes of models, learning paradigms, evaluation measures, defining interpretability. motivating factors. Main requirements.

(2) Interpretability of rule-based models

Classes of rule-based architectures. Modularity properties. Design principles. A taxonomy of interpretation mechanisms: decomposition of conditions, granular conditions and conclusions, symbolic descriptors, linguistic approximation.

(3) Tree structures in machine learning

Decision trees- a generic construct and design strategies. From decision trees to rules. Interpretation mechanisms. Generalized architectures and learning: random forests and gradient boosting in the development of decision trees.

(4) Logic networks and induction of concepts

Logic expressions and logic neurons. Conditional (context) clustering. Concept formation through clustering. Concept refinement via gradient-based learning of logic networks and their interpretation.

(5) Federated learning

Privacy and security requirements. Federated learning of rules. Performance evaluation. Three-tier federated learning and granular rules.

Lecturer: Witold Pedrycz

Date/Time: 9:00 – 10:00am, 8th Oct-5th Nov, 2021

Online Platform: Tencent Conference ID: 329 6526 8237

Brief introduction of the lecturer: 

Witold pedrycz, Chair Professor at the University of Alberta, Canada, academician of the Royal Canadian Academy of Sciences, foreign academician of the Polish Academy of Sciences, and member of the association of electrical and electronic engineers. He is editor in chief of Information Sciences and wires data mining and knowledge discovery, and co editor in chief of int. J. of granular computing and J. of data information and management. Published 18 monographs, h index 109. His main research fields include computational intelligence, fuzzy modeling and particle computing, knowledge discovery and data science, pattern recognition, data science, knowledge-based neural networks and control engineering.