Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
478691 | European Journal of Operational Research | 2010 | 8 Pages |
Abstract
In a new learning paradigm called Induction over Strategic Agents, the principal anticipates possible alteration of attributes by agents wishing to achieve a positive classification. In many cases, agents are constrained on how an attribute can be modified. For example, attribute values may have upper and lower bounds or they may need to belong to a certain set of possible values such as binary valued attributes like “pays bills on time” or be linearly dependent like the relationships between accounting entries in an income statement. In this paper, we explore Induction over Strategic Agents for a class of problems where attributes are binary values.
Related Topics
Physical Sciences and Engineering
Computer Science
Computer Science (General)
Authors
Fidan Boylu, Haldun Aytug, Gary J. Koehler,