CS7637: Knowledge-Based AI (Cognitive Systems)

This course explored how to build AI agents that reason the way humans do — using explicit knowledge representations like semantic networks, frames, production rules, and case-based reasoning rather than purely statistical learning — applied to classic cognitive-modeling problems.

Raven’s Progressive Matrices (RPM) Agent

The course’s semester-long capstone: building an AI agent that solves visual analogy puzzles from Raven’s Progressive Matrices, a real human IQ-test format, by reasoning over pixel-level and shape-level transformations between images. Built incrementally across milestones to a final agent tested against a large bank of problem sets.

Mini-Projects & Homework Agents

A series of smaller named agents, each implementing a different classical AI reasoning technique: a block-world planning agent, case-based and rule-based classification and diagnosis agents over a toy “monster” domain, and semantic-network and sentence-parsing agents for knowledge representation.