Accession Number:

ADP007160

Title:

Genetic Optimization for Exploratory Projection Pursuit,

Descriptive Note:

Corporate Author:

ADVANCED DECISION SYSTEMS MOUNTAIN VIEW CA

Personal Author(s):

• Crawford, Stuart L.

Report Date:

1992-01-01

Pagination or Media Count:

4.0

Abstract:

Exploratory Projection Pursuit is a technique for forming projections of a multivariate point cloud and searching for those projections that reveal the most structure. The search component is typically some variant of a steepest descent procedure and, particularly when the search space is ill-behaved, leaves open the possibility that the best projection will not be found. Genetic Algorithms are generally applicable optimization techniques, well suited for search spaces in which more traditional techniques fail. This paper describes experiments designed to ascertain the effectiveness of Genetic Algorithms as optimizers for exploratory projection pursuit.

Descriptors:

• *ALGORITHMS
• *GENETICS
• CLOUDS
• DESCENT
• OPTIMIZATION
• PAPER
• SEARCHING
• STRUCTURES

Subject Categories:

• Statistics and Probability

Distribution Statement:

APPROVED FOR PUBLIC RELEASE