Advanced search options

Advanced Search Options 🞨

Browse by author name (“Author name starts with…”).

Find ETDs with:

in
/  
in
/  
in
/  
in

Written in Published in Earliest date Latest date

Sorted by

Results per page:

You searched for +publisher:"Georgia Tech" +contributor:("Joyner, David"). One record found.

Search Limiters

Last 2 Years | English Only

No search limiters apply to these results.

▼ Search Limiters


Georgia Tech

1. Shegheva, Snejana. A computational model for solving raven’s progressive matrices intelligence test.

Degree: MS, Computer Science, 2018, Georgia Tech

Graphical models offer techniques for capturing the structure of many problems in real- world domains and provide means for representation, interpretation, and inference. The modeling framework provides tools for discovering rules for solving problems by exploring structural relationships. We present the Structural Affinity method that uses graphical models for first learning and subsequently recognizing the pattern for solving problems on the Raven’s Progressive Matrices Test of general human intelligence. Recently there has been considerable work on computational models of addressing the Raven’s test using various representations ranging from fractals to symbolic structures. In contrast, our method uses Markov Random Fields parameterized by affinity factors to discover the structure in the geometric problems and induce the rules of Carpenter et al.’s cognitive model of problem-solving on the Raven’s Progressive Matrices Test. We provide a computational account that first learns the structure of Raven’s problem and then predicts the solution by computing the probability of the correct answer by recognizing patterns corresponding to Carpenter et al.’s rules. We demonstrate that the performance of our model on the Standard Raven Progressive Matrices is comparable with existing state of the art models. In this report, we raise and attempt to address research questions about the knowledge representation that provides a sufficient opportunity to capture a pattern in geometrical intelligence tests such as Raven Progressive Matrices. We show how a minimal representation facilitates pattern extraction process by proposing a method for organizing the representational units using the framework of probabilistic graphical models. By orchestrating techniques from mathematics, data science and computer science, we design an agent that can explain its responses while still predicting accurate solutions. And finally, we discuss the key takeaways about knowledge representation, structure discoveries, heuristic reasoning, and a possible connection to the cognitive thinking process. Advisors/Committee Members: Goel, Ashok (committee member), McGreggor, Keith (committee member), Joyner, David (committee member).

Subjects/Keywords: Probabilistic graphical models; Machine learning

Record DetailsSimilar RecordsGoogle PlusoneFacebookTwitterCiteULikeMendeleyreddit

APA · Chicago · MLA · Vancouver · CSE | Export to Zotero / EndNote / Reference Manager

APA (6th Edition):

Shegheva, S. (2018). A computational model for solving raven’s progressive matrices intelligence test. (Masters Thesis). Georgia Tech. Retrieved from http://hdl.handle.net/1853/60191

Chicago Manual of Style (16th Edition):

Shegheva, Snejana. “A computational model for solving raven’s progressive matrices intelligence test.” 2018. Masters Thesis, Georgia Tech. Accessed August 21, 2019. http://hdl.handle.net/1853/60191.

MLA Handbook (7th Edition):

Shegheva, Snejana. “A computational model for solving raven’s progressive matrices intelligence test.” 2018. Web. 21 Aug 2019.

Vancouver:

Shegheva S. A computational model for solving raven’s progressive matrices intelligence test. [Internet] [Masters thesis]. Georgia Tech; 2018. [cited 2019 Aug 21]. Available from: http://hdl.handle.net/1853/60191.

Council of Science Editors:

Shegheva S. A computational model for solving raven’s progressive matrices intelligence test. [Masters Thesis]. Georgia Tech; 2018. Available from: http://hdl.handle.net/1853/60191

.