Overview

Description

CALENDAR DESCRIPTION:

Intelligent systems are knowledge-based computer programs which emulate the reasoning abilities of human experts. This introductory course will analyze the underlying artificial intelligence methodology and survey advances in rule-based systems, constraint solving, incremental reasoning, intelligent backtracking and heuristic local search methods. We will look specifically at research applications in intelligent scheduling, configuration and planning. The course is intended for graduate students with a reasonable background in symbolic programming.

COURSE DETAILS:

Many real computing applications involve computational problems for which no generally efficient algorithms are known. These problems often involve relatively small inputs, but astronomically-sized search spaces in which solutions may be well hidden. Developing practical software for solving these problems is challenging and costly. This course examines methods for automatically solving these problems based on high-level specifications, much as we use a query language rather than low-level algorithms to answer complex questions about a large data set. The course will pursue two tracks in parallel. In one, we will examine the design and use of real systems that are products of major companies or widely used research tools, including hands-on experience. In the other, we will develop formal foundations to support principled design and analysis of such systems and their use in critical applications, and to understand their strengths and limitations. We will begin with lectures on fundamentals, and toward the end shift to considering current issues and research directions. The course is suitable for students with an interest in research or in tackling problems that are not well-solved. A strong discrete math background is recommended.

Topics

• Intractable search and optimization problems
• Combinatorial and heuristic search
• Best-first search, A* search, bounded-suboptimal search, advanced search techniques
• A case study of the multi-agent pathfinding problem
• (Selective) applications in planning, multi-agent/robot systems, games, etc.

Grading