One of the most interesting and challenging problems to confront those interested in the learning process in recent years is the entire area of the self-instructional device, or "teaching machine." The idea of the teaching machine is not new, for Pressey (49) in 1926 wrote concerning a device he had developed, and at the same time indicated that he had had such a device in mind for "a number of years." After this introduction by Pressey, the teaching machine movement lay dormant for several years with only an occasional article written that had any direct relationship to this area. This was not to last indefinitely, however, because during the past ten years the interest has gradually been growing to the point that at the present time this movement demands consideration. It is difficult to identify precisely why this has been the case, but a few reasons might be suggested. The demand on education is greater now than it has ever been before (47, 57): there are more people wanting education, there are more students receiving education, the percentage of school-age persons participating is increasing, and the teacher-pupil ratio is not remaining at a desirable level. To further intensify this problem, much more is being demanded. from education in the general areas of curriculum and desired levels of competence. Since this presents the educational system with the obvious task of keeping abreast of these demands, the educator has been forced to search for more effective and efficient methods of instruction. Glaser (28) in his review suggests that the trend is toward closer cooperation and coordination of effort between "educational psychology" and the "science of learning." The experimentalist and the learning theorist are working more closely together on training and learning problems than they ever have before. A final reason for this increased interest, according to Holland (31), is that in the past the interest has been largely on the device itself, but in recent years this has shifted to focus upon the fact that a person's behavior can be altered in situations outside of the laboratory by the application and utilization of certain psychological principles. These same principles can be incorporated in the teaching machine. Not only is this movement intriguing, but it presents a great challenge, for there are a great number of problems, first to be identified, and second to be solved. From all indications this interest will not dissipate, but rather will become more universal with widespread implications for the student, the teacher, the administrator, the psychologist, and the parent (4, 18, 47). The implications are not confined to t hose associated with a school setting, for as Skinner (57) indicates there is additional application in home study, industrial training, military training, and special education of the exceptional individual. No doubt there are others but this will serve to illustrate the potentially wide-spread effects.

In this thesis, we focus on studying the performance of general regularized discriminant analysis (RDA) classifiers. The data used for analysis is assumed to follow Gaussian mixture model with different means and covariances. RDA offers a rich class of regularization options, covering as special cases the regularized linear discriminant analysis (RLDA) and the regularized quadratic discriminant analysis (RQDA) classi ers. We analyze RDA under the double asymptotic regime where the data dimension and the training size both increase in a proportional way. This double asymptotic regime allows for application of fundamental results from random matrix theory. Under the double asymptotic regime and some mild assumptions, we show that the asymptotic classification error converges to a deterministic quantity that only depends on the data statistical parameters and dimensions. This result not only implicates some mathematical relations between the misclassification error and the class statistics, but also can be leveraged to select the optimal parameters that minimize the classification error, thus yielding the optimal classifier. Validation results on the synthetic data show a good accuracy of our theoretical findings. We also construct a general consistent estimator to approximate the true classification error in consideration of the unknown previous statistics. We benchmark the performance of our proposed consistent estimator against classical estimator on synthetic data. The observations demonstrate that the general estimator outperforms others in terms of mean squared error (MSE).

Abstract not available

This thesis is concerned with the development of virtual Human-Machine Interfaces (HMI) and navigation method for wheelchair systems. For virtual HMI, hand gesture recognition is employed, and two different hand gesture recognition algorithms have been developed. One is based on the geometric properties of a hand shape, and the other algorithm is based on the curvature of a hand shape contour. In the hand gesture recognition algorithm using geometric properties of a hand shape, eight non-dimensional parameters are computed and identifies hand shapes by comparing the ranges of the parameters to the statistical range information. This algorithm is invariant at scale, but does not work properly if the forearm of a hand shape is cluttered. The curvature based hand gesture recognition algorithm recognizes hand gestures using a combination of hand shape contour geometry and a non-dimensional quantity derived using the curvatures of the hand shape contour. The algorithm produces a set of signatures of the contour and identifies each hand gesture by finding matched template signatures. This algorithm is not affected by the forearm of a hand shape, but the scaling procedure is required. The developed gesture recognition system is implemented on a wheelchair in two different modes of operations, namely, the manual mode and the map (autonomous) mode. In the manual mode, the user continuously interacts with the wheelchair and controls the speed and the steering using the position and the orientation of hand gestures. In the map mode, the user selects a desired destination by pointing with a hand gesture onto a known map, and then the wheelchair initiates autonomous navigation. For wheelchair navigation, a doorway recognition algorithm and an obstacle avoidance algorithm have been developed. The wheelchair is localised by finding the doorway template in the specified zone. If the doorway recognition algorithm does not detect the doorway, it navigates to find the doorway using the obstacle avoidance algorithm. The obstacle avoidance algorithm finds obstacle edge points using range data and decides a safe passage for wheelchair navigation to find the doorway. Results obtained by implementing the above mentioned algorithms are presented.

The Dairy industry requires accurate detection of oestrus(heat) in dairy cows to maximise output of the animals. Traditionally this is a process dependant on human observation and interpretation of the various signs of heat. Many areas of the dairy industry can be automated, however the detection of oestrus is an area that still requires human experts. This thesis investigates the application of Machine Learning classification techniques, on dairy cow milking data provided by the Livestock Improvement Corporation, to predict oestrus. The usefulness of various ensemble learning algorithms such as Bagging and Boosting are explored as well as specific skewed data techniques. An empirical study into the effectiveness of classifiers designed to target skewed data is included as a significant part of the investigation. Roughly Balanced Bagging and the novel Under Bagging classifiers are explored in considerable detail and found to perform quite favourably over the SMOTE technique for the datasets selected. This study uses non-dairy, commonplace, Machine Learning datasets; many of which are found in the UCI Machine Learning Repository.

Knowledge Based Systems (KBS) have been actively investigated since the early period of AI. There are four common methods of building expert systems: modeling approaches, programming approaches, case-based approaches and machine-learning approaches. One particular technique is Ripple Down Rules (RDR) which may be classified as an incremental case-based approach. Knowledge needs to be acquired from experts in the context of individual cases viewed by them. In the RDR framework, the expert adds a new rule based on the context of an individual case. This task is simple and only affects the expert???s workflow minimally. The rule added fixes an incorrect interpretation made by the KBS but with minimal impact on the KBS's previous correct performance. This provides incremental improvement. Despite these strengths of RDR, there are some limitations including rule redundancy, lack of intermediate features and lack of models. This thesis addresses these RDR limitations by applying automatic learning algorithms to reorganize the knowledge base, to learn intermediate features and possibly to discover domain models. The redundancy problem occurs because rules created in particular contexts which should have more general application. We address this limitation by reorganizing the knowledge base and removing redundant rules. Removal of redundant rules should also reduce the number of future knowledge acquisition sessions. Intermediate features improve modularity, because the expert can deal with features in groups rather than individually. In addition to the manual creation of intermediate features for RDR, we propose the automated discovery of intermediate features to speed up the knowledge acquisition process by generalizing existing rules. Finally, the Ripple Down Rules approach facilitates rapid knowledge acquisition as it can be initialized with a minimal ontology. Despite minimal modeling, we propose that a more developed knowledge model can be extracted from an existing RDR KBS. This may be useful in using RDR KBS for other applications. The most useful of these three developments was the automated discovery of intermediate features. This made a significant difference to the number of knowledge acquisition sessions required.

In bread-making, the baking process is one of the key steps to produce the final product quality attributes including texture, color and flavor, as a result of several thermal reactions such as non-enzymatic browning reaction, starch gelatinisation and protein denaturation. These thermal reactions are dominated by heat and mass transfer mechanisms inside an oven chamber as well as inside the dough pieces. In this study, an industrial baking process was divided into 4 zones. Experiments were conducted, and mathematical models were developed to account for the heat and mass contribution as well as their consequent impacts on the product qualities. Monitoring systems were developed and installed inside an industrial oven to evaluate oven performance, including temperature profile and airflow pattern. Many other tests and experiments were conducted and results given in some detail. To deal with the complexity of a continuous baking process, a three dimensional transient-state CFD model with moving grids was established to account for the effect of oven load on heat transfer in the oven chamber. The dynamic response of the travelling tin temperature profiles could be predicted in accordance with a change in the oven load. The modelled tin temperature profiles showed a good agreement with the measured tin temperature profiles from the actual industrial baking process. Finally, the three-dimensional CFD model could provide guidance in manipulating the oven condition to achieve the optimum temperature profile in the industrial travelling-tray baking oven. / Doctor of Philosophy (PhD)

Classification learning is dominated by systems which induce large numbers of small axis-orthogonal decision surfaces. This strongly biases such systems towards particular hypothesis types but there is reason believe that many domains have underlying concepts which do not involve axis orthogonal surfaces. Further, the multiplicity of small decision regions mitigates against any holistic appreciation of the theories produced by these systems, notwithstanding the fact that many of the small regions are individually comprehensible. This thesis investigates modeling concepts as large geometric structures in n-dimensional space. Convex hulls are a superset of the set of axis orthogonal hyperrectangles into which axis orthogonal systems partition the instance space. In consequence, there is reason to believe that convex hulls might provide a more flexible and general learning bias than axis orthogonal regions. The formation of convex hulls around a group of points of the same class is shown to be a usable generalisation and is more general than generalisations produced by axis-orthogonal based classifiers, without constructive induction, like decision trees, decision lists and rules. The use of a small number of large hulls as a concept representation is shown to provide classification performance which can be better than that of classifiers which use a large number of small fragmentary regions for each concept. A convex hull based classifier, CH1, has been implemented and tested. CH1 can handle categorical and continuous data. Algorithms for two basic generalisation operations on hulls, inflation and facet deletion, are presented. The two operations are shown to improve the accuracy of the classifier and provide moderate classification accuracy over a representative selection of typical, largely or wholly continuous valued machine learning tasks. The classifier exhibits superior performance to well-known axis-orthogonal-based classifiers when presented with domains where the underlying decision surfaces are not axis parallel. The strengths and weaknesses of the system are identified. One particular advantage is the ability of the system to model domains with approximately the same number of structures as there are underlying concepts. This leads to the possibility of extraction of higher level mathematical descriptions of the induced concepts, using the techniques of computational geometry, which is not possible from a multiplicity of small regions.

Thesis (Ph.D.)--McMaster University, 2005. / Supervisors: M.A. Elbestawi, G.M. Bone. Includes bibliographical references (leaves 117-122).

Thesis (Ph. D.)--Washington State University, December 2007. / Includes bibliographical references (p. 98-107).