Thesis (M.S.)--Tennessee Technological University, 2009. / Title from title page screen (viewed on Feb. 24, 2010). Bibliography: leaves 78-83.

Thesis (Ph. D.)--University of New South Wales, 2004. / Also available online.

Thesis (Ph.D.)--Australian National University, 2004.

Thesis (MCIS - Computer and Information Sciences) -- AUT University, 2009. / Includes bibliographical references. Also held in print (xiii, 96 leaves : ill., charts, graphs ; 30 cm.) in the Archive at the City Campus (T 006.31 LIA)

Thesis (M.S.)--Oregon State University, 2000. / Typescript (photocopy). Includes bibliographical references (leaf 21). Also available on the World Wide Web.

Thesis (M.Sc.)--University of Alberta, 2010. / Title from PDF file main screen (viewed on July 8, 2010). A thesis submitted to the Faculty of Graduate Studies and Research in partial fulfillment of the requirements for the degree of Master of Science, Department of Computing Science, University of Alberta. Includes bibliographical references.

Thesis (Ph. D.)--Oregon State University, 2008. / Printout. Includes bibliographical references (leaves 133-141). Also available on the World Wide Web.

The success of many computer vision problems and machine learning algorithms critically depends on the quality of the chosen distance metrics or similarity functions. Due to the fact that the real-data at hand is inherently task- and data-dependent, learning an appropriate distance metric or similarity function from data for each specific task is usually superior to the default Euclidean distance or cosine similarity. This thesis mainly focuses on developing new metric and similarity learning models for three tasks: unconstrained face verification, person re-identification and kNN classification. Unconstrained face verification is a binary matching problem, the target of which is to predict whether two images/videos are from the same person or not. Concurrently, person re-identification handles pedestrian matching and ranking across non-overlapping camera views. Both vision problems are very challenging because of the large transformation differences in images or videos caused by pose, expression, occlusion, problematic lighting and viewpoint. To address the above concerns, two novel methods are proposed. Firstly, we introduce a new dimensionality reduction method called Intra-PCA by considering the robustness to large transformation differences. We show that Intra-PCA significantly outperforms the classic dimensionality reduction methods (e.g. PCA and LDA). Secondly, we propose a novel regularization framework called Sub-SML to learn distance metrics and similarity functions for unconstrained face verifica- tion and person re-identification. The main novelty of our formulation is to incorporate both the robustness of Intra-PCA to large transformation variations and the discriminative power of metric and similarity learning, a property that most existing methods do not hold. Working with the task of kNN classification which relies a distance metric to identify the nearest neighbors, we revisit some popular existing methods for metric learning and develop a general formulation called DMLp for learning a distance metric from data. To obtain the optimal solution, a gradient-based optimization algorithm is proposed which only needs the computation of the largest eigenvector of a matrix per iteration. Although there is a large number of studies devoted to metric/similarity learning based on different objective functions, few studies address the generalization analysis of such methods. We describe a novel approch for generalization analysis of metric/similarity learning which can deal with general matrix regularization terms including the Frobenius norm, sparse L1-norm, mixed (2, 1)-norm and trace-norm. The novel models developed in this thesis are evaluated on four challenging databases: the Labeled Faces in the Wild dataset for unconstrained face verification in still images; the YouTube Faces database for video-based face verification in the wild; the Viewpoint Invariant Pedestrian Recognition database for person re-identification; the UCI datasets for kNN classification. Experimental results show that the proposed methods yield competitive or state-of-the-art performance.

Thesis (Ph. D.)--Lehigh University, 2005. / Includes vita. Includes bibliographical references (leaves 163-177).

Computer-based interactive learning environments in physics can help students to differentiate between their intuitive views on natural phenomena and the formalisms of Newtonian physics. This thesis describes empirical investigations of a specific type of interactive learning environments, computer-based simulations. In many cases computer simulations deal with a simplified and idealised version of the natural phenomenon. Presenting the user with a simplification of reality is seen as one of the advantages of simulations, since too complex and too realistic simulations may sometimes be overwhelming for learners and may not permit the identification of the underlying model. Yet implications arise about the degree to which students either expect or perceive simulations to be real and how these expectations and perceptions affect their interaction with the simulation. Reality for the purposes of this research is considered to be a construct comprising the visual fidelity (fidelity) and the complexity of the underlying physical model (complexity) of the simulation. Evaluation of a number of simulations, two case studies and interviews with simulation designers and educators suggested these components. Altering the relation between fidelity and complexity levels affects students' learning and contributes to the students' perception of reality. This is demonstrated in a study of a number of simulations of the same physical phenomenon (Newtonian collisions) with degrees of fidelity and complexity which have been examined to test this hypothesis. Two empirical studies were then conducted to investigate the use of simulations which represented different fidelity and complexity levels. Analyses were carried out on videotapes and questionnaires of students interacting collaboratively with the simulations (40 hours of computer based activity). The empirical approaches to these studies, reports on work done, including the emerging data in multiple forms (questionnaires, video and audio tapes of the students interaction) and its analysis are presented in this thesis. The work reported looks at students' interaction with the simulations (pre to post test learning gain and issues concerning pre and post testing), their comments on the interface and the model underlying the simulation. The thesis supports the view that well designed computer-based simulations can promote learning and that design issues are essential to the creation of successful simulations. The findings claim that: a) enhanced fidelity of an instructional simulation has positive effects on the learner outcome, b) interfaces which use multiple representations offer valuable information which facilitates problem solving strategies, and c) low complexity simulations are better suited to novice learners. These outcomes are presented as implications for simulation design and the use and development of a syntax in simulation design is also discussed (design criteria for how systems might be built). Finally the outcomes' applicability, the limitations of the studies, as well as the scope for further research that should lead to an understanding of the factors which promote successful use of simulations in the teaching of physics are presented.