The efforts addressed in this thesis refer to applying nonlinear risk predictive techniques based on logistic regression to medical diagnostic test data. This study is motivated and pursued to address the following: 1. To extend logistic regression model of biostatistics to medical informatics 2. Computational preemptive and predictive testing to determine the probability of occurrence (p) of an event by fitting a data set to a (logit function) logistic curve: Finding upper and lower bounds on p based on stochastical considerations 3. Using the model developed on available (clinical) data to illustrate the bounds-limited performance of the prediction. Relevant analytical methods, computational efforts and simulated results are presented. Using the results compiled, the risk evaluation in medical diagnostics is discussed with real-world examples. Conclusions are enumerated and inferences are made with directions for future studies. / by Alice Horn Dupont. / Thesis (M.S.C.S.)--Florida Atlantic University, 2011. / Includes bibliography. / Electronic reproduction. Boca Raton, Fla., 2011. Mode of access: World Wide Web.

by Kwok-Fai Cheung. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1999. / Includes bibliographical references (leaves 100-103). / Abstracts in English and Chinese. / Chapter 1 --- Introduction --- p.4 / Chapter 1.1 --- Summary of Contributions --- p.5 / Chapter 1.2 --- Organization of the Thesis --- p.6 / Chapter 2 --- Current Methods and Problems --- p.8 / Chapter 2.1 --- Statisticians --- p.8 / Chapter 2.1.1 --- ARMA --- p.8 / Chapter 2.1.1.1 --- Moving Average models --- p.8 / Chapter 2.1.1.2 --- Autoregressive models --- p.9 / Chapter 2.1.1.3 --- Stationary Process --- p.10 / Chapter 2.1.1.4 --- Autoregressive-Moving Average model --- p.10 / Chapter 2.1.1.5 --- Parameter Estimation --- p.11 / Chapter 2.2 --- Financial Researchers --- p.11 / Chapter 2.2.1 --- Efficient Market Theory --- p.11 / Chapter 2.3 --- Computer Scientists --- p.12 / Chapter 2.3.1 --- Expert System --- p.12 / Chapter 2.3.2 --- Neural Network --- p.14 / Chapter 2.3.2.1 --- Multilayer Perceptron --- p.14 / Chapter 2.3.2.2 --- Radial Basis Function Network (RBF) --- p.19 / Chapter 2.4 --- Research Apart from Prediction and Trading in Finance --- p.22 / Chapter 2.4.1 --- Derivatives Valuation and Hedging --- p.22 / Chapter 2.4.1.1 --- Volatility --- p.22 / Chapter 2.4.2 --- Pricing of Initial Public Offering --- p.24 / Chapter 2.4.3 --- Credit Rating --- p.25 / Chapter 2.4.4 --- Financial Health Assessment --- p.26 / Chapter 2.5 --- Discussion --- p.27 / Chapter 3 --- Correlation Basis Function Network --- p.28 / Chapter 3.1 --- Formulation of CBF network --- p.31 / Chapter 3.2 --- First Order Learning Algorithm --- p.32 / Chapter 3.3 --- Summary --- p.35 / Chapter 4 --- Applications of CBF Network in Stock trading --- p.36 / Chapter 4.1 --- Data Representation --- p.36 / Chapter 4.2 --- Data Pre-processing --- p.38 / Chapter 4.2.1 --- Input data pre-processing --- p.38 / Chapter 4.2.2 --- Output data pre-processing --- p.38 / Chapter 4.3 --- Multiple CBF Networks for Generation of Trading Signals --- p.41 / Chapter 4.4 --- Output Data Post-processing --- p.41 / Chapter 4.5 --- Trader's Interpretation --- p.43 / Chapter 4.6 --- Maximum profit trading system --- p.45 / Chapter 4.7 --- Performance Evaluation --- p.46 / Chapter 5 --- Applications of CBF Network in Warrant trading --- p.48 / Chapter 5.1 --- Option Model --- p.48 / Chapter 5.2 --- Warrant Model --- p.49 / Chapter 5.3 --- Black-Scholes Pricing Formula --- p.51 / Chapter 5.4 --- Using CBF Network for choosing warrants --- p.53 / Chapter 5.5 --- Trading System --- p.53 / Chapter 5.5.1 --- Trading System by Black-Scholes Model --- p.54 / Chapter 5.5.2 --- Trading System by Warrant Sensitivity --- p.55 / Chapter 5.6 --- Learning of Parameters in Warrant Sensitivity Model by Hierarchi- cal CBF Network --- p.57 / Chapter 5.7 --- Experimental Results --- p.59 / Chapter 5.7.1 --- Aggregate profit --- p.62 / Chapter 5.8 --- Summary and Discussion --- p.69 / Chapter 6 --- Analysis of CBF Network and other models --- p.72 / Chapter 6.1 --- Time and Space Complexity --- p.72 / Chapter 6.1.1 --- RBF Network --- p.72 / Chapter 6.1.2 --- CBF Network --- p.74 / Chapter 6.1.3 --- Black-Scholes Pricing Formula --- p.74 / Chapter 6.1.4 --- Warrant Sensitivity Model --- p.75 / Chapter 6.2 --- "Model Confidence, Prediction Confidence and Model Stability" --- p.76 / Chapter 6.2.1 --- Model and Prediction Confidence --- p.77 / Chapter 6.2.2 --- Model Stability --- p.77 / Chapter 6.2.3 --- Linear Model Analysis --- p.79 / Chapter 6.2.4 --- CBF Network Analysis --- p.82 / Chapter 6.2.5 --- Black-Scholes Pricing Formula Analysis --- p.84 / Chapter 7 --- Conclusion --- p.93 / Chapter 7.1 --- Neural Network and Statistical Modeling --- p.95 / Chapter 7.2 --- Financial Markets --- p.95 / Chapter A --- RBF Network Parameters Estimation --- p.101 / Chapter A.1 --- Least Squares --- p.101 / Chapter A.2 --- Gradient Descent Algorithm --- p.103 / Chapter B --- Further study on Black-Scholes Model --- p.104

Kei-Keung Hung. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1999. / Includes bibliographical references (leaves 127-128). / Abstract also in Chinese. / Abstract --- p.i / Acknowledgments --- p.iii / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Static Portfolio Techniques --- p.1 / Chapter 1.2 --- Neural Network Approach --- p.2 / Chapter 1.3 --- Contributions of this Thesis --- p.3 / Chapter 1.4 --- Application of this Research --- p.4 / Chapter 1.5 --- Organization of this Thesis --- p.4 / Chapter 2 --- Literature Review --- p.6 / Chapter 2.1 --- Standard Markowian Portfolio Optimization (SMPO) and Sharpe Ratio --- p.6 / Chapter 2.2 --- Downside Risk --- p.9 / Chapter 2.3 --- Augmented Lagrangian Method --- p.10 / Chapter 2.4 --- Adaptive Supervised Learning Decision (ASLD) System --- p.13 / Chapter I --- Static Portfolio Optimization Techniques --- p.19 / Chapter 3 --- Modified Portfolio Sharpe Ratio Maximization (MPSRM) --- p.20 / Chapter 3.1 --- Experiment Setting --- p.21 / Chapter 3.2 --- Downside Risk and Upside Volatility --- p.22 / Chapter 3.3 --- Investment Diversification --- p.24 / Chapter 3.4 --- Analysis of the Parameters H and B of MPSRM --- p.27 / Chapter 3.5 --- Risk Minimization with Control of Expected Return --- p.30 / Chapter 3.6 --- Return Maximization with Control of Expected Downside Risk --- p.32 / Chapter 4 --- Variations of Modified Portfolio Sharpe Ratio Maximization --- p.34 / Chapter 4.1 --- Soft-max Version of Modified Portfolio Sharpe Ratio Maximization (SMP- SRM) --- p.35 / Chapter 4.1.1 --- Applying Soft-max Technique to Modified Portfolio Sharpe Ratio Maximization (SMPSRM) --- p.35 / Chapter 4.1.2 --- Risk Minimization with Control of Expected Return --- p.37 / Chapter 4.1.3 --- Return Maximization with Control of Expected Downside Risk --- p.38 / Chapter 4.2 --- Soft-max Version of MPSRM with Entropy-like Regularization Term (SMPSRM-E) --- p.39 / Chapter 4.2.1 --- Using Entropy-like Regularization term in Soft-max version of Modified Portfolio Sharpe Ratio Maximization (SMPSRM-E) --- p.39 / Chapter 4.2.2 --- Risk Minimization with Control of Expected Return --- p.41 / Chapter 4.2.3 --- Return Maximization with Control of Expected Downside Risk --- p.43 / Chapter 4.3 --- Analysis of Parameters in SMPSRM and SMPSRM-E --- p.44 / Chapter II --- Neural Network Approach --- p.48 / Chapter 5 --- Investment on a Foreign Exchange Market using ASLD system --- p.49 / Chapter 5.1 --- Investment on A Foreign Exchange Portfolio --- p.49 / Chapter 5.2 --- Two Important Issues Revisited --- p.51 / Chapter 6 --- Investment on Stock market using ASLD System --- p.54 / Chapter 6.1 --- Investment on Hong Kong Hang Seng Index --- p.54 / Chapter 6.1.1 --- Performance of the Original ASLD System --- p.54 / Chapter 6.1.2 --- Performances After Adding Several Heuristic Strategies --- p.55 / Chapter 6.2 --- Investment on Six Different Stock Indexes --- p.61 / Chapter 6.2.1 --- Structure and Operation of the New System --- p.62 / Chapter 6.2.2 --- Experimental Results --- p.63 / Chapter III --- Combination of Static Portfolio Optimization techniques with Neural Network Approach --- p.67 / Chapter 7 --- Combining the ASLD system with Different Portfolio Optimizations --- p.68 / Chapter 7.1 --- Structure and Operation of the New System --- p.69 / Chapter 7.2 --- Combined with the Standard Markowian Portfolio Optimization (SMPO) --- p.70 / Chapter 7.3 --- Combined with the Modified Portfolio Sharpe Ratio Maximization (MP- SRM) --- p.72 / Chapter 7.4 --- Combined with the MPSRM ´ؤ Risk Minimization with Control of Ex- pected Return --- p.74 / Chapter 7.5 --- Combined with the MPSRM ´ؤ Return Maximization with Control of Expected Downside Risk --- p.76 / Chapter 7.6 --- Combined with the Soft-max Version of MPSRM (SMPSRM) --- p.77 / Chapter 7.7 --- Combined with the SMPSRM - Risk Minimization with Control of Ex- pected Return --- p.79 / Chapter 7.8 --- Combined with the SMPSRM ´ؤ Return Maximization with Control of Expected Downside Risk --- p.80 / Chapter 7.9 --- Combined with the Soft-max Version of MPSRM with Entropy-like Reg- ularization Term (SMPSRM-E) --- p.82 / Chapter 7.10 --- Combined with the SMPSRM-E ´ؤ Risk Minimization with Control of Expected Return --- p.84 / Chapter 7.11 --- Combined with the SMPSRM-E ´ؤ Return Maximization with Control of Expected Downside Risk --- p.86 / Chapter IV --- Software Developed --- p.93 / Chapter 8 --- Windows Application Developed --- p.94 / Chapter 8.1 --- Decision on Platform and Programming Language --- p.94 / Chapter 8.2 --- System Design --- p.96 / Chapter 8.3 --- Operation of our program --- p.97 / Chapter 9 --- Conclusion --- p.103 / Chapter A --- Algorithm for Portfolio Sharpe Ratio Maximization (PSRM) --- p.105 / Chapter B --- Algorithm for Improved Portfolio Sharpe Ratio Maximization (ISRM) --- p.107 / Chapter C --- Proof of Regularization Term Y --- p.109 / Chapter D --- Algorithm for Modified Portfolio Sharpe Ratio Maximization (MP- SRM) --- p.111 / Chapter E --- Algorithm for MPSRM with Control of Expected Return --- p.113 / Chapter F --- Algorithm for MPSRM with Control of Expected Downside Risk --- p.115 / Chapter G --- Algorithm for SMPSRM with Control of Expected Return --- p.117 / Chapter H --- Algorithm for SMPSRM with Control of Expected Downside Risk --- p.119 / Chapter I --- Proof of Entropy-like Regularization Term --- p.121 / Chapter J --- Algorithm for SMPSRM-E with Control of Expected Return --- p.123 / Chapter K --- Algorithm for SMPSRM-E with Control of Expected Downside Riskl25 Bibliography --- p.127

Ding, Ling. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2009. / Includes bibliographical references (leaves [43]-45). / Abstract also in Chinese. / Thesis Committee --- p.i / Abstract --- p.ii / Acknowledgement --- p.iv / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Background --- p.1 / Chapter 1.2 --- Motivation --- p.3 / Chapter 1.3 --- Thesis Organization --- p.4 / Chapter 2 --- Representative Skylines and Basic Properties --- p.6 / Chapter 2.1 --- Existing Formulation --- p.6 / Chapter 2.1.1 --- Max-dominance Representative Skyline --- p.6 / Chapter 2.1.2 --- Defects of the Existing Formulation --- p.7 / Chapter 2.2 --- Our Formulation --- p.9 / Chapter 2.2.1 --- Distance-based Representative Skyline --- p.9 / Chapter 2.2.2 --- Properties of Our Formulation --- p.10 / Chapter 2.3 --- Problem Definition --- p.12 / Chapter 3 --- The Two-dimensional Case --- p.13 / Chapter 3.1 --- Algorithm 2D-opt --- p.13 / Chapter 3.2 --- Time Complexity --- p.15 / Chapter 3.3 --- Computing Covering Circles --- p.15 / Chapter 4 --- The Higher-dimensional Case --- p.18 / Chapter 4.1 --- NP-hardness and 2-approximation --- p.18 / Chapter 4.1.1 --- Proof of NP-hardness --- p.18 / Chapter 4.1.2 --- Algorithm naive-greedy --- p.19 / Chapter 4.2 --- Algorithm I-greedy --- p.20 / Chapter 4.2.1 --- Conservative Skyline --- p.22 / Chapter 4.2.2 --- Access Order --- p.23 / Chapter 4.3 --- Computing the Maximum Representative Distance --- p.27 / Chapter 5 --- Experiments --- p.30 / Chapter 5.1 --- Data --- p.30 / Chapter 5.2 --- Representation Quality --- p.31 / Chapter 5.2.1 --- Representative Skylines --- p.31 / Chapter 5.2.2 --- Representation Error Comparison --- p.33 / Chapter 5.3 --- Efficiency --- p.34 / Chapter 5.3.1 --- Running Time Comparison --- p.34 / Chapter 5.3.2 --- Scalability Comparison --- p.37 / Chapter 6 --- Related Work --- p.39 / Chapter 7 --- Conclusions --- p.41 / A List of Publications --- p.42 / Bibliography --- p.43

Tsue, Wing Yeung. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2008. / Includes bibliographical references (leaves 94-100). / Abstracts in English and Chinese. / Abstract --- p.i / Acknowledgement --- p.vii / Chapter 1 --- Introduction --- p.1 / Chapter 2 --- Reinforcement Learning (RL) --- p.7 / Chapter 2.1 --- Objective of RL --- p.8 / Chapter 2.2 --- Algorithms in RL --- p.9 / Chapter 2.2.1 --- Dynamic Programming --- p.9 / Chapter 2.2.2 --- Monte Carlo Methods --- p.11 / Chapter 2.2.3 --- Temporal-Difference Learning and Q-Learning --- p.12 / Chapter 2.3 --- Example: Maze --- p.13 / Chapter 2.4 --- Artificial Neural Network to Approximate Q-Function --- p.14 / Chapter 2.5 --- Literatures on Trading a Single Asset by RL --- p.16 / Chapter 2.6 --- Literatures on Portfolio Management by RL --- p.19 / Chapter 2.7 --- Summary --- p.20 / Chapter 3 --- Portfolio Management (PM) --- p.21 / Chapter 3.1 --- Buy-and-Hold Strategy --- p.22 / Chapter 3.2 --- Mean-Variance Analysis --- p.23 / Chapter 3.3 --- Constant Rebalancing Algorithm --- p.24 / Chapter 3.4 --- Universal Portfolio Algorithm --- p.25 / Chapter 3.5 --- ANTI COR Algorithm --- p.26 / Chapter 4 --- PM on RL Traders --- p.30 / Chapter 4.1 --- Implementation of Single-Asset RL Traders --- p.32 / Chapter 4.1.1 --- State Formation --- p.32 / Chapter 4.1.2 --- Actions and Immediate Reward --- p.38 / Chapter 4.1.3 --- Update --- p.38 / Chapter 4.2 --- Experiments --- p.41 / Chapter 4.3 --- Discussion --- p.47 / Chapter 5 --- RL-Bascd Portfolio Management (RLPM) --- p.49 / Chapter 5.1 --- Overview --- p.52 / Chapter 5.2 --- Two-Asset RL System --- p.54 / Chapter 5.2.1 --- State Formation --- p.55 / Chapter 5.2.2 --- Action --- p.61 / Chapter 5.2.3 --- Update Rule --- p.64 / Chapter 5.3 --- Portfolio Construction --- p.67 / Chapter 5.4 --- Choice of Window Size w --- p.70 / Chapter 5.5 --- Empirical Results --- p.73 / Chapter 5.5.1 --- "Effect of Window Size w on 1 Layer of RLPMw, and 2 Layers of RLPMW" --- p.76 / Chapter 5.5.2 --- Comparing RLPM to Other Strategies --- p.80 / Chapter 5.5.3 --- Effect of Transaction Cost A on RLPMw --- p.83 / Chapter 6 --- Conclusion --- p.89 / Bibliography --- p.94

Modern civil commercial transport aircraft provide the means for the safest of all forms of transportation. While advanced computer technology ranging from flight management computers to warning and alerting devices contributed to flight safety significantly, it is undisputed that the flightcrew represents the most frequent primary cause factor in airline accidents. From a system perspective, machine actors such as the autopilot and human actors (the flightcrew) try to achieve goals (desired states of the aircraft). The set of activities to achieve a goal is called a function. In modern flightdecks both machine actors and human actors perform functions. Recent accident studies suggest that deficiencies in the flightcrew's ability to monitor how well either machines or themselves perform a function are a factor in many accidents and incidents. As humans are inherently bad monitors, this study proposes a method to automatically assess the status of a function in order to increase flight safety as part of an intelligent pilot aid, called the AgendaManager. The method was implemented for the capture altitude function: seeking to attain and maintain a target altitude. Fuzzy systems were used to compute outputs indicating how well the capture altitude function was performed from inputs describing the state of the aircraft. In order to conform to human expert assessments, the fuzzy systems were trained using a genetic algorithm (GA) whose objective was to minimize the discrepancy between system outputs and human expert assessments based on 72 scenarios. The resulting systems were validated by analyzing how well they conformed to new data drawn from another 32 scenarios. The results of the study indicated that even though the training procedure facilitated by the GA was able to improve conformance to human expert assessments, overall the systems performed too poorly to be deployed in a real environment. Nevertheless, experience and insights gained from the study will be valuable in the development of future automated systems to perform function assessment. / Graduation date: 1998

Modern commercial air travel is considered by most transportation specialists to be the safest of all forms of transportation. While safe, any loss of life is tragic and the only really acceptable state of commercial air transport safety is that of 'zero-tolerance' where no accident is acceptable. Research has demonstrated that the largest single causal component for airline accidents is the flightcrew. In addition, the recent automation of these machines has created many new safety concerns involving flightcrew situational awareness, human-machine interfacing, workload, attention, and complexity, to name a few. These concerns led to a series of studies developing, refining, and testing numerous aspects of this issue. The studies incorporated ASRS (Aviation Safety Reporting System) incident reports, NTSB (National Transportation Safety Board) and other accident reports, and the development of a CTM (cockpit task management) system. The information gained from this research led to the development of an agent-based cockpit task aiding system termed the AgendaManager. A traditional text-based display similar to that used in the CTMS study was developed, optimized, and integrated with existing systems like EICAS (Engine Indicating and Crew Alerting System) using a visual display development guide developed from a literature review. An alternate display incorporating graphics and located on the primary flight display (PFD) was also developed in an effort to improve pilot agenda management performance. Both of the interfaces were developed using a comprehensive visual display design guide compiled through a literature review. The Agenda Manager displays were tested in order to determine if the PFD enhancements improved agenda management performance. Eleven general aviation pilots participated in the study, three in the pilot study and eight in the main study. Results from the main study indicate little, if any, difference in agenda management performance in regards to the display format used. In general, the study demonstrated the usefulness of the display guidelines, importance of tracking instrument rating when using general aviation pilots in an experiment, and the equivalence of 'round-trip' scenarios. / Graduation date: 1998

This thesis is concerned with the problem of how to make decisions in an uncertain world. We use a model of uncertainty based on Markov decision problems, and develop a number of algorithms for decision-making both for the planning problem, in which the model is known in advance, and for the reinforcement learning problem in which the decision-making agent does not know the model and must learn to make good decisions by trial and error. The basis for much of this work is the use of structural representations of problems. If a problem is represented in a structured way we can compute or learn plans that take advantage of this structure for computational gains. This is because the structure allows us to perform abstraction. Rather than reasoning about each situation in which a decision must be made individually, abstraction allows us to group situations together and reason about a whole set of them in a single step. Our approach to abstraction has the additional advantage that we can dynamically change the level of abstraction, splitting a group of situations in two if they need to be reasoned about separately to find an acceptable plan, or merging two groups together if they no longer need to be distinguished. We present two planning algorithms and one learning algorithm that use this approach. A second idea we present in this thesis is a novel approach to the exploration problem in reinforcement learning. The problem is to select actions to perform given that we would like good performance now and in the future. We can select the current best action to perform, but this may prevent us from discovering that another action is better, or we can take an exploratory action, but we risk performing poorly now as a result. Our Bayesian approach makes this tradeoff explicit by representing our uncertainty about the values of states and using this measure of uncertainty to estimate the value of the information we could gain by performing each action. We present both model-free and model-based reinforcement learning algorithms that make use of this exploration technique. Finally, we show how these ideas fit together to produce a reinforcement learning algorithm that uses structure to represent both the problem being solved and the plan it learns, and that selects actions to perform in order to learn using our Bayesian approach to exploration.

Wu, Xiaobing. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2009. / Includes bibliographical references (leaves [48]-51). / Abstract also in Chinese. / Thesis Committee --- p.i / Abstract --- p.ii / Acknowledgement --- p.iv / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Motivation --- p.1 / Chapter 1.2 --- Thesis Organization --- p.7 / Chapter 2 --- Related Work --- p.9 / Chapter 2.1 --- Static Skyline --- p.9 / Chapter 2.2 --- Monochromatic Reverse Skyline --- p.11 / Chapter 2.3 --- Other Skyline-Related Topics --- p.13 / Chapter 2.4 --- Bichromatic Reverse Nearest Neighbor --- p.14 / Chapter 3 --- Basic Properties of Bichromatic Reverse Skyline --- p.16 / Chapter 3.1 --- Bichromatic Reverse Skyline --- p.16 / Chapter 3.2 --- Basic Properly --- p.19 / Chapter 4 --- The BRS Algorithm --- p.24 / Chapter 4.1 --- Midway Conversion --- p.24 / Chapter 4.2 --- Heuristics --- p.27 / Chapter 4.3 --- The Algorithm --- p.32 / Chapter 5 --- Any Mixture of Bi- and Uni-directional Dimensions --- p.37 / Chapter 6 --- Experiments --- p.40 / Chapter 6.1 --- Bi-directional Retrieval --- p.40 / Chapter 6.2 --- Mixture of Bi- and Uni-directional Axes --- p.44 / Chapter 7 --- Conclusions --- p.46 / Chapter A --- List of Publications --- p.47 / Bibliography --- p.48

Dissertation submitted in complete fulfilment for the requirements of the Degree of Master of Technology in Information Technology, M L Sultan Technikon, 1998. / Decision Support Systems (DSS) are interactive computer-based systems developed to support managers in complex tasks requiring human judgment. DSS utilise data, provide an easy user interface and allow for the decision maker's own insights. / M