Date of Award

12-1-2012

Document Type

Campus Access Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computer Science

First Advisor

Marc Pomplun

Second Advisor

Dan Simovici

Third Advisor

Jun Suzuki

Abstract

My dissertation focuses on developing computational models of eye movements for understanding how cognitive processes (e.g., visual information processing, word recognition, attention, and oculomotor control) can work together to perform a complex everyday task (e.g., reading or scene viewing). In a theoretical framework, many biologically-inspired computational methods were used and found psychologically plausible to predict human behaviors and simulate human cognition. For eye movements in reading, I proposed models of visual encoding, word identification, and semantic integration in contexts. Using singular value decomposition (SVD), I was able to predict the most important strokes for Chinese character recognition (Wang, Angele, Schotter, Yang, Simovici, Pomplun, & Rayner, under revision). Furthermore, I used a vector space model (latent semantic analysis, LSA) to explain how readers rate the semantic transparency of English and Chinese compound words. A linear regression model was then developed to estimate contextual predictability during reading (Wang, Chen, Ko, Pomplun, & Rayner, 2010), and a connectionist model was used to represent the activations of concepts in working memory (Plummer, Wang, Tzeng, Pomplun, & Rayner, 2012).

My interests in reading and vision studies provided interdisciplinary research opportunities, which I pursued by applying methods and concepts from reading research to the viewing of real-world scenes. Regarding eye movements in natural scene viewing, I studied when and where we fixate, resulting in a model for gaze transition using LSA (Wang, Hwang, & Pomplun, 2010; Hwang, Wang, & Pomplun, 2011). The final part of my dissertation focuses on studying how texts attract attention in natural scene viewing (Wang & Pomplun, 2012) compared to attraction by saliency and edge density. I have also developed a model of this effect of texts on visual attention that includes an automatic text detector (Wang, Lu, Lim, & Pomplun, 2012).

The results of my doctoral thesis will broaden our understanding of low-level and higher-level cognitive processing as well as cultural differences during reading and real-world scene viewing. The findings should eventually lead to practical applications, e.g., contribute to the development of more effective automatic text detectors, or making a great difference to visually challenged people's lives by assisting them in reading and scene viewing.

Comments

Free and open access to this Campus Access Dissertation is made available to the UMass Boston community by ScholarWorks at UMass Boston. Those not on campus and those without a UMass Boston campus username and password may gain access to this dissertation through resources like Proquest Dissertations & Theses Global or through Interlibrary Loan. If you have a UMass Boston campus username and password and would like to download this work from off-campus, click on the "Off-Campus UMass Boston Users" link above.

Share

COinS