Seminar on Computational Learning and Adaptation

  From generic learning tasks to knowledge discovery process management
An 18 year journey
[slides]

Yoram Reich
Center for Design Research, Stanford University
on leave from Faculty of Engineering, Tel Aviv University

Many real problems are complex and require multiple perspectives and methods for their solution. Using complex knowledge discovery techniques to address these problems is equally complex and often requires exercising multiple discovery methods in an intricate process. This talk describes my experience in addressing such complex discovery problems. Starting with the concept of generic learning tasks that defines discovery functionality, I arrive at a specification of process information management for knowledge discovery. I will describe examples of knowledge discovery/machine learning systems that illustrate these ideas.  These include a system for bridge design, a conversational case-based system for customer relationship management, a lesson-learned case-based system with weak knowledge, and a system that learns complex rules from medical and process related databases.

Return to the seminar schedule