CS 6768 - Bridging Reasoning and Learning

Fall. 3 credits. Student option grading.

Prerequisite: CS 4700 and CS 4780. Enrollment limited to: graduated students or permission of instructor. Only offered Fall of every other year.

K. Ellis.

Studies AI problems that bridge machine learning and automated reasoning, and hybrid techniques for addressing those problems. Concrete topics include inductive and abductive reasoning; statistical-relational learning; combining perception with symbolic processing; and applying learning to combinatorial search and probabilistic inference. Themes of the course include (1) structure and compositionality, both in neural networks and in symbolic systems; and (2) probabilistic framings as an interface between discrete and continuous representations. Includes homework assignments and a final project. Students should be comfortable with artificial intelligence at the level of (CS 4700) and machine learning (at the level of CS 4780).

Outcome 1: Analyze AI problems by decomposing them into learning and reasoning components.

Outcome 2: Demonstrate facility in engineering hybrid discrete/continuous AI systems.

Outcome 3: Understand open problems in the field and be equipped to work on them.