Alzheimer’s disease is typically detected using a combination of cognitive-behavioral assessment exams and interviews of both the patient and a family member or caregiver, both administered and interpreted by a trained physician. This procedure, while standard in medical practice, can be time consuming and expensive for both the patient and the diagnostician especially because proper training is required to interpret the collected information and determine an appropriate diagnosis. The use of machine learning techniques to augment diagnostic procedures has been previously examined in limited capacity but to date no research examines real-world medical applications of predictive analytics for health records and cognitive exam scores. This dissertation seeks to examine the efficacy of detecting cognitive impairment due to Alzheimer’s disease using machine learning, including multi-modal neural network architectures, with a real-world clinical dataset used to determine the accuracy and applicability of the generated models. An in-depth analysis of each type of data (e.g. cognitive exams, questionnaires, demographics) as well as the cognitive domains examined (e.g. memory, attention, language) is performed to identify the most useful targets, with cognitive exams and questionnaires being found to be the most useful features and short-term memory, attention, and language found to be the most important cognitive domains. In an effort to reduce medical costs and streamline procedures, optimally predictive and efficient groups of features were identified and selected, with the best performing and economical group containing only three questions and one cognitive exam component, producing an accuracy of 85%. The most effective diagnostic scoring procedure was examined, with simple threshold counting based on medical documentation being identified as the most useful. Overall predictive analysis found that Alzheimer’s disease can be detected most accurately using a bimodal multi-input neural network model using separated cognitive domains and questionnaires, with a detection accuracy of 88% using the real-world testing set, and that the technique of analyzing domains separately serves to significantly improve model efficacy compared to models that combine them. / Includes bibliography. / Dissertation (Ph.D.)--Florida Atlantic University, 2019. / FAU Electronic Theses and Dissertations Collection

<div> <div> <div> <p>The objective of this study is that combination of molecular dynamics (MD) simulations and machine learning to complement each other. In this study, four steps are conducted. </p> <p>First is based on the empirical potentials development in silicon nanowires for theory parts of molecular dynamics. Many-body empirical potentials have been developed for the last three decades, and with the advance of supercomputers, these potentials are expected to be even more useful for the next three decades. Atomistic calculations using empirical potentials can be particularly useful in understanding the structural aspects of Si or Si-H systems, however, existing empirical potentials have many errors of parameters. We propose a novel technique to understand and construct interatomic potentials with an emphasis on parameter fitting, in which the relationship between material properties and potential parameters is explained. The input database has been obtained from density functional theory (DFT) calculations with the Vienna ab initio simulation package (VASP) using the projector augmented-wave method within the generalized gradient approximation. The DFT data are used in the fitting process to guarantee the compatibility within the context of multiscale modeling. </p> <p>Second, application part of MD simulations, enhancement of mechanical properties was focused in this research by using MEAM potentials. For instance, Young’s modulus, ultimate tensile strength, true strain, true stress and stress-strain relationship were calculated for nanosized Cu-precipitates using quenching & partitioning (Q&P) processing and nanosized Fe3C strengthened ultrafine-grained (UFG) ferritic steel. In the stress-strain relationship, the structure of simulation is defined using the constant total number of particles, constant-energy, constant-volume ensemble (NVE) is pulled in the y-direction, or perpendicular to the boundary interface, to increase strain. The strain in increased for a specified number of times in a loop and the stress is calculated at each point before the simulation loops.</p></div></div> </div> <div> <div> <div> <p>Third, based on the MD simulations, machine learning and the peridynamics are applied to prediction of disk damage patterns. The peridynamics is the nonlocal extension of classical continuum mechanics and same as MD model. Especially, FEM is based on the partial differential equations, however, partial derivatives do not exist on crack and damage surfaces. To complement this problem, the peridynamics was used which is based on the integral equations and overcome deficiencies in the modeling of deformation discontinuities. In this study, the forward problem (i), if we have images of damage and crack, crack patterns are predicted by using trained data compared to true solutions which are hit by changing the x and y hitting coordinates on the disk. The inverse problem (ii), if we have images of damage and crack, the corresponding hitting location, indenter velocity and indenter size are predicted by using trained data. Furthermore, we did the regression analysis for the images of the crack patterns with Neural processes to predict the crack patterns. In the regression problem, by representing the results of the variance according to the epochs, it can be confirmed that the result of the variance is decreased by increasing the epoch through the neural processes. Therefore, the result of the training gradually improves, and the ranges of the variance are expressed as 0 to 0.035. The most critical point of this study is that the neural processes makes an accurate prediction even if the information of the training data is missing or not enough. The results show that if the context points are set to 10, 100, 300, and 784, the training information is deliberately omitted such as context points of 10, 100 and 300, and the predictions are different when context points are significantly lower. However, when comparing the results of context points 100 and 784, the predicted results appear to be very similar to each other because of the Gaussian processes in the neural processes. Therefore, if the training data is trained through the Neural processes, the missing information of training data can be supplemented to predict the results. </p> <p>Finally, we predicted the data by applying various data using deep learning as well as MD simulation data. This study applied the deep learning to Cryo-EM images and Line Trip (LT) data with power systems. In this study, deep learning method was applied to reduce the effort of selection of high-quality particles. This study proposes a learning frame structure using deep learning and aims at freeing passively selecting high quality particles as the ultimate goal. For predicting the line trip data and bad data detection, we choose to analyze the frequency signal because suddenly the frequency changes in the power system due to events such as generator trip, line trip or load shedding in large power systems. </p> </div> </div> </div>

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

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.

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

Reinforcement Learning (RL) is the study of agents that learn optimal behavior by interacting with and receiving rewards and punishments from an unknown environment. RL agents typically do this by learning value functions that assign a value to each state (situation) or to each state-action pair. Recently, there has been a growing interest in using hierarchical methods to cope with the complexity that arises due to the huge number of states found in most interesting real-world problems. Hierarchical methods seek to reduce this complexity by the use of temporal and state abstraction. Like most RL methods, most hierarchical RL methods optimize the discounted total reward that the agent receives. However, in many domains, the proper criteria to optimize is the average reward per time step. In this thesis, we adapt the concepts of hierarchical and recursive optimality, which are used to describe the kind of optimality achieved by hierarchical methods, to the average reward setting and show that they coincide under a condition called Result Distribution Invariance. We present two new model-based hierarchical RL methods, HH-learning and HAH-learning, that are intended to optimize the average reward. HH-learning is a hierarchical extension of the model-based, average-reward RL method, H-learning. Like H-learning, HH-learning requires exploration in order to learn correct domain models and optimal value function. HH-learning can be used with any exploration strategy whereas HAH-learning uses the principle of "optimism under uncertainty", which gives it a built-in "auto-exploratory" feature. We also give the hierarchical and auto-exploratory hierarchical versions of R-learning, a model-free average reward method, and a hierarchical version of ARTDP, a model-based discounted total reward method. We compare the performance of the "flat" and hierarchical methods in the task of scheduling an Automated Guided Vehicle (AGV) in a variety of settings. The results show that hierarchical methods can take advantage of temporal and state abstraction and converge in fewer steps than the flat methods. The exception is the hierarchical version of ARTDP. We give an explanation for this anomaly. Auto-exploratory hierarchical methods are faster than the hierarchical methods with ��-greedy exploration. Finally, hierarchical model-based methods are faster than hierarchical model-free methods. / Graduation date: 2003

Developing logic in machines has always been an area of concern for scientists. Automated Theorem Proving is a field that has implemented the concept of logical consequence to a certain level. However, if the number of available axioms is very large then the probability of getting a proof for a conjecture in a reasonable time limit can be very small. This is where the ability to learn from previously proved theorems comes into play. If we see in our own lives, whenever a new situation S(NEW) is encountered we try to recollect all old scenarios S(OLD) in our neural system similar to the new one. Based on them we then try to find a solution for S(NEW) with the help of all related facts F(OLD) to S(OLD). Similar is the concept in this research. The thesis deals with developing a solution and finally implementing it in a tool that tries to prove a failed conjecture (a problem that the ATP system failed to prove) by extracting a sufficient set of axioms (we call it Refined Axiom Set (RAS)) from a large pool of available axioms. The process is carried out by measuring the similarity of a failed conjecture with solved theorems (already proved) of the same domain. We call it "process1", which is based on syntactic selection of axioms. After process1, RAS may still have irrelevant axioms, which motivated us to apply semantic selection approach on RAS so as to refine it to a much finer level. We call this approach as "process2". We then try to prove failed conjecture either from the output of process1 or process2, depending upon whichever approach is selected by the user. As for our testing result domain, we picked all FOF problems from the TPTP problem domain called SWC, which consisted of 24 broken conjectures (problems for which the ATP system is able to show that proof exists but not able to find it because of limited resources), 124 failed conjectures and 274 solved theorems. The results are produced by keeping in account both the broken and failed problems. The percentage of broken conjectures being solved with respect to the failed conjectures is obviously higher and the tool has shown a success of 100 % on the broken set and 19.5 % on the failed ones.