Can Mekik, RPI Graduate Student

 

Can Mekik, RPI Graduate Student

Sage 4101

October 25, 2017 12:00 PM - 1:30 PM

Human inspired models of psychometric intelligence tests offer a way to better understand the nature of intelligence, both natural and artificial. In this talk, I present a new model of an influential intelligence test, the Raven's Progressive Matrices, and report results of a computational experiment testing the model. The model employs a sub-symbolic similarity-based approach to the task and represents an analysis of Raven's Matrices as a relative entropy minimization problem.

Implications will be discussed for sub-symbolic approaches to generality in artificial intelligence and for the development of new test items in the style of Raven's Matrices.

 

 

Add to calendar
Share|