Lan Weiyao. / "December 2003." / Thesis (Ph.D.)--Chinese University of Hong Kong, 2003. / Includes bibliographical references (p. 163-172). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Electronic reproduction. Ann Arbor, MI : ProQuest Information and Learning Company, [200-] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Mode of access: World Wide Web. / Abstracts in English and Chinese.

This first part of this thesis concerns the discovery and parameter determination of three hot Jupiter planets, first detected with by the SuperWASP collaboration, and their planetary nature is confirmed with the modelling of radial velocity measurements and further ground-based transit lightcurves. WASP-92b, WASP-93b and WASP-118b are all hot Jupiters with short orbital periods – 2.17, 2.73 and 4.05 days respectively. The analysis in this thesis finds WASP-92b to have R[sub]p = 1.461 ± 0.077 R[sub]J and M[sub]p = 0.805 ± 0.068 M[sub]J; WASP-93b to have R[sub]p = 1.597 ± 0.077 R[sub]J and M[sub]p = 1.47 ± 0.029 M[sub]J, and WASP-118b to have R[sub]p = 1.440 ± 0.036 R[sub]J and M[sub]p = 0.514 ± 0.020 M[sub]J. The second part of this thesis presents three novel approaches to modelling the effect of stellar limb darkening when fitting exoplanet transit lightcurves. The first method trains a Gaussian Process to interpolate between pre-calculated limb darkening coefficients for the non-linear limb darkening law. The method uses existing knowledge of the stellar atmosphere parameters as the constraints of the determined limb darkening coefficients for the host star of the transiting exoplanet system. The second method deploys an artificial neural network to model limb darkening without the requirement of a parametric approximation of the form of the limb profile. The neural network is trained for a specific bandpass directly from the outputs of stellar atmosphere models, allowing predictions to be made for the stellar intensity at a given position on the stellar surface for values of the T[sub]eff , log g and [Fe/H]. The efficacy of the method is demonstrated by accurately fitting a transit lightcurve for the transit of Venus, and for a single transit lightcurve of TRES-2b. The final limb darkening modelling method proposes an adjustment to the neural network model to account for the fact that the stellar radius is not constant across wavelengths. The method also allows the full variation in light at the edge of the star to be modelled by not assuming a sharp boundary at the limb.

The enormous increase in the production of electronic data in today's information era has led to more challenges in analysing and understanding of the data. The rise in the innovations of technology devices, computers and the Internet has made it much easier to collect and store different kind of data ranging from personal, medical, financial, and scientific data. The growth in the amount of the generated data has introduced the term “Big Data” to describe this extremely high-dimensional and yet complex data. Making sense of the generated data sets is of great importance for the discovery of meaningful information that can be used to support decision making. Data mining techniques have been designed as a process for ex-ploring these data sets to extract meaning for decision making. An essential phase of the data mining procedure is the data transformation that involves the selection of input parameters. Selecting the right input parameters has a great impact on the performance of machine learning algorithms. Currently, there are existing manual statistical methods that are used for this task, but these are difficult to use, time consuming and require an expert. Automated data analysis is the initial step to relieve this burden from humans, through the provision of a systematic procedure of inspecting, transforming and modelling data for knowledge discovery. This project presents a novel method that exploits the power of self-organization for a sys-tematic procedure of conducting and inspecting data classification, with the identification of input parameters that are important for the process. The developed method can be used on different classification problems with practical application in various areas such as health con-dition monitoring in health care, machinery fault detection and analysis, and financial instrument analysis among others.

by Po-Ming Wong. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1997. / Includes bibliographical references (leaves 94-100). / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Introduction --- p.1 / Chapter 1.2 --- Image Data --- p.2 / Chapter 1.2.1 --- Feature Detection --- p.2 / Chapter 1.3 --- Neural Networks --- p.4 / Chapter 1.4 --- Invariant Object Recognition --- p.5 / Chapter 1.5 --- Thesis Outline --- p.7 / Chapter 2 --- Feature Extraction --- p.8 / Chapter 2.1 --- Review of the Principle Component Analysis (PCA) Method --- p.9 / Chapter 2.1.1 --- Theory --- p.10 / Chapter 2.2 --- Covariance Operator --- p.13 / Chapter 2.3 --- Corner Extraction Method --- p.16 / Chapter 2.3.1 --- Corner Detection on the Surface of an Object --- p.16 / Chapter 2.3.2 --- Corner Detection at Boundary Region --- p.17 / Chapter 2.3.3 --- Steps in Corner Detection Process --- p.21 / Chapter 2.4 --- Experiment Results and Discussion --- p.23 / Chapter 2.4.1 --- Features Localization --- p.27 / Chapter 2.4.2 --- Preparing Feature Points for Matching Process --- p.32 / Chapter 2.5 --- Summary --- p.32 / Chapter 3 --- Invariant Graph Matching Using High-Order Hopfield Network --- p.36 / Chapter 3.1 --- Review of the Hopfield Network --- p.37 / Chapter 3.1.1 --- 3D Image Matching Algorithm --- p.40 / Chapter 3.1.2 --- Iteration Algorithm --- p.44 / Chapter 3.2 --- Third-order Hopfield Network --- p.45 / Chapter 3.3 --- Experimental Results --- p.49 / Chapter 3.4 --- Summary --- p.58 / Chapter 4 --- Hopfield Network for 2D and 3D Mirror-Symmetric Image Match- ing --- p.59 / Chapter 4.1 --- Introduction --- p.59 / Chapter 4.2 --- Geometric Symmetry --- p.60 / Chapter 4.3 --- Motivation --- p.62 / Chapter 4.4 --- Third-order Hopfield Network for Solving 2D Symmetry Problems --- p.66 / Chapter 4.5 --- Forth-order Hopfield Network for Solving 3D Symmetry Problem --- p.71 / Chapter 4.6 --- Experiment Results --- p.78 / Chapter 4.7 --- Summary --- p.88 / Chapter 5 --- Conclusion --- p.90 / Chapter 5.1 --- Results and Contributions --- p.90 / Chapter 5.2 --- Future Work --- p.92 / Bibliography --- p.94

by Lee Shing-yan. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2001. / Includes bibliographical references (leaves 126-133). / Abstracts in English and Chinese. / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Problem Statement --- p.4 / Chapter 1.2 --- Contributions --- p.4 / Chapter 1.3 --- Thesis Organization --- p.5 / Chapter 2 --- Background --- p.7 / Chapter 2.1 --- Bayesian Networks --- p.7 / Chapter 2.1.1 --- A Simple Example [42] --- p.8 / Chapter 2.1.2 --- Formal Description and Notations --- p.9 / Chapter 2.1.3 --- Learning Bayesian Network from Data --- p.14 / Chapter 2.1.4 --- Inference on Bayesian Networks --- p.18 / Chapter 2.1.5 --- Applications of Bayesian Networks --- p.19 / Chapter 2.2 --- Bayesian Network Classifiers --- p.20 / Chapter 2.2.1 --- The Classification Problem in General --- p.20 / Chapter 2.2.2 --- Bayesian Classifiers --- p.21 / Chapter 2.2.3 --- Bayesian Network Classifiers --- p.22 / Chapter 2.3 --- Evolutionary Computation --- p.28 / Chapter 2.3.1 --- Four Kinds of Evolutionary Computation --- p.29 / Chapter 2.3.2 --- Cooperative Coevolution --- p.31 / Chapter 3 --- Bayesian Network Learning Algorithms --- p.33 / Chapter 3.1 --- Related Work --- p.34 / Chapter 3.1.1 --- Using GA --- p.34 / Chapter 3.1.2 --- Using EP --- p.36 / Chapter 3.1.3 --- Criticism of the Previous Approaches --- p.37 / Chapter 3.2 --- Two New Strategies --- p.38 / Chapter 3.2.1 --- A Hybrid Framework --- p.38 / Chapter 3.2.2 --- A New Operator --- p.39 / Chapter 3.3 --- CCGA --- p.44 / Chapter 3.3.1 --- The Algorithm --- p.45 / Chapter 3.3.2 --- CI Test Phase --- p.46 / Chapter 3.3.3 --- Cooperative Coevolution Search Phase --- p.47 / Chapter 3.4 --- HEP --- p.52 / Chapter 3.4.1 --- A Novel Realization of the Hybrid Framework --- p.54 / Chapter 3.4.2 --- Merging in HEP --- p.55 / Chapter 3.4.3 --- Prevention of Cycle Formation --- p.55 / Chapter 3.5 --- Summary --- p.56 / Chapter 4 --- Evaluation of Proposed Learning Algorithms --- p.57 / Chapter 4.1 --- Experimental Methodology --- p.57 / Chapter 4.2 --- Comparing the Learning Algorithms --- p.61 / Chapter 4.2.1 --- Comparing CCGA with MDLEP --- p.63 / Chapter 4.2.2 --- Comparing HEP with MDLEP --- p.65 / Chapter 4.2.3 --- Comparing CCGA with HEP --- p.68 / Chapter 4.3 --- Performance Analysis of CCGA --- p.70 / Chapter 4.3.1 --- Effect of Different α --- p.70 / Chapter 4.3.2 --- Effect of Different Population Sizes --- p.72 / Chapter 4.3.3 --- Effect of Varying Crossover and Mutation Probabilities --- p.73 / Chapter 4.3.4 --- Effect of Varying Belief Factor --- p.76 / Chapter 4.4 --- Performance Analysis of HEP --- p.77 / Chapter 4.4.1 --- The Hybrid Framework and the Merge Operator --- p.77 / Chapter 4.4.2 --- Effect of Different Population Sizes --- p.80 / Chapter 4.4.3 --- Effect of Different --- p.81 / Chapter 4.4.4 --- Efficiency of the Merge Operator --- p.84 / Chapter 4.5 --- Summary --- p.85 / Chapter 5 --- Learning Bayesian Network Classifiers --- p.87 / Chapter 5.1 --- Issues in Learning Bayesian Network Classifiers --- p.88 / Chapter 5.2 --- The Multinet Classifier --- p.89 / Chapter 5.3 --- The Augmented Bayesian Network Classifier --- p.91 / Chapter 5.4 --- Experimental Methodology --- p.94 / Chapter 5.5 --- Experimental Results --- p.97 / Chapter 5.6 --- Discussion --- p.103 / Chapter 5.7 --- Application in Direct Marketing --- p.106 / Chapter 5.7.1 --- The Direct Marketing Problem --- p.106 / Chapter 5.7.2 --- Response Models --- p.108 / Chapter 5.7.3 --- Experiment --- p.109 / Chapter 5.8 --- Summary --- p.115 / Chapter 6 --- Conclusion --- p.116 / Chapter 6.1 --- Summary --- p.116 / Chapter 6.2 --- Future Work --- p.118 / Chapter A --- A Supplementary Parameter Study --- p.120 / Chapter A.1 --- Study on CCGA --- p.120 / Chapter A.1.1 --- Effect of Different α --- p.120 / Chapter A.1.2 --- Effect of Different Population Sizes --- p.121 / Chapter A.1.3 --- Effect of Varying Crossover and Mutation Probabilities --- p.121 / Chapter A.1.4 --- Effect of Varying Belief Factor --- p.122 / Chapter A.2 --- Study on HEP --- p.123 / Chapter A.2.1 --- The Hybrid Framework and the Merge Operator --- p.123 / Chapter A.2.2 --- Effect of Different Population Sizes --- p.124 / Chapter A.2.3 --- Effect of Different Δα --- p.124 / Chapter A.2.4 --- Efficiency of the Merge Operator --- p.125

Although distinguishing different sounds in noisy environment is a relative easy task for human, source separation has long been extremely difficult in audio signal processing. The problem is challenging for three reasons: the large variety of sound type, the abundant mixing conditions and the unclear mechanism to distinguish sources, especially for similar sounds. In recent years, the neural network based methods achieved impressive successes in various problems, including the speech enhancement, where the task is to separate the clean speech out of the noise mixture. However, the current deep learning based source separator does not perform well on real recorded noisy speech, and more importantly, is not applicable in a more general source separation scenario such as overlapped speech. In this thesis, we firstly propose extensions for the current mask learning network, for the problem of speech enhancement, to fix the scale mismatch problem which is usually occurred in real recording audio. We solve this problem by combining two additional restoration layers in the existing mask learning network. We also proposed a residual learning architecture for the speech enhancement, further improving the network generalization under different recording conditions. We evaluate the proposed speech enhancement models on CHiME 3 data. Without retraining the acoustic model, the best bi-direction LSTM with residue connections yields 25.13% relative WER reduction on real data and 34.03% WER on simulated data. Then we propose a novel neural network based model called “deep clustering” for more general source separation tasks. We train a deep network to assign contrastive embedding vectors to each time-frequency region of the spectrogram in order to implicitly predict the segmentation labels of the target spectrogram from the input mixtures. This yields a deep network-based analogue to spectral clustering, in that the embeddings form a low-rank pairwise affinity matrix that approximates the ideal affinity matrix, while enabling much faster performance. At test time, the clustering step “decodes” the segmentation implicit in the embeddings by optimizing K-means with respect to the unknown assignments. Experiments on single channel mixtures from multiple speakers show that a speaker-independent model trained on two-speaker and three speakers mixtures can improve signal quality for mixtures of held-out speakers by an average over 10dB. We then propose an extension for deep clustering named “deep attractor” network that allows the system to perform efficient end-to-end training. In the proposed model, attractor points for each source are firstly created the acoustic signals which pull together the time-frequency bins corresponding to each source by finding the centroids of the sources in the embedding space, which are subsequently used to determine the similarity of each bin in the mixture to each source. The network is then trained to minimize the reconstruction error of each source by optimizing the embeddings. We showed that this frame work can achieve even better results. Lastly, we introduce two applications of the proposed models, in singing voice separation and the smart hearing aid device. For the former, a multi-task architecture is proposed, which combines the deep clustering and the classification based network. And a new state of the art separation result was achieved, where the signal to noise ratio was improved by 11.1dB on music and 7.9dB on singing voice. In the application of smart hearing aid device, we combine the neural decoding with the separation network. The system firstly decodes the user’s attention, which is further used to guide the separator for the targeting source. Both objective study and subjective study show the proposed system can accurately decode the attention and significantly improve the user experience.

This Thesis surveys the landscape of Data Augmentation for image datasets. Completing this survey inspired further study into a method of generative modeling known as Generative Adversarial Networks (GANs). A survey on GANs was conducted to understood recent developments and the problems related to training them. Following this survey, four experiments were proposed to test the application of GANs for data augmentation and to contribute to the quality improvement in GAN-generated data. Experimental results demonstrate the effectiveness of GAN-generated data as a pre-training metric. The other experiments discuss important characteristics of GAN models such as the refining of prior information, transferring generative models from large datasets to small data, and automating the design of Deep Neural Networks within the context of the GAN framework. This Thesis will provide readers with a complete introduction to Data Augmentation and Generative Adversarial Networks, as well as insights into the future of these techniques. / Includes bibliography. / Thesis (M.S.)--Florida Atlantic University, 2019. / FAU Electronic Theses and Dissertations Collection

Today transportation systems are facing big transitions all over the world. We created fly overs, roads under the ground, bridges over the river and ocean to get efficient access and to increase the road connectivity. Our transportation system is more intelligent than ever. Our traffic signaling system became adaptive. Our vehicles equipped with new gadgets and we developed new tools for more efficient analysis of traffic. Our research relies on existing traffic infrastructure to generate better understanding of traffic. More specifically, this research focused on traffic and UAV cameras to extract information about the traffic. Our first goal was to create an automatic system to count the cars using traffic cameras. To achieve this goal, we implemented Background Subtraction Method (BSM) and OverFeat Framework. BSM compares consecutive frames to detect the moving objects. Because BSM only works for ideal lab conditions, therefor we implemented a Convolutional Neural Network (CNN) based classification algorithm called OverFeat Framework. We created different segments on the road in various lanes to tabulate the number of passing cars. We achieved 96.55% accuracy for car counting irrespective of different visibility conditions of the day and night. Our second goal was to find out traffic density. We implemented two CNN based algorithms: Single Shot Detection (SSD) and MobileNet-SSD for vehicle detection. These algorithms are object detection algorithms. We used traffic cameras to detect vehicles on the roads. We utilized road markers and light pole distances to determine distances on the road. Using the distance and count information we calculated density. SSD is a more resource intense algorithm and it achieved 92.97% accuracy. MobileNet-SSD is a lighter algorithm and it achieved 79.30% accuracy. Finally, from a moving platform we estimated the velocity of multiple vehicles. There are a lot of roads where traffic cameras are not available, also traffic monitoring is necessary for special events. We implemented Faster R-CNN as a detection algorithm and Discriminative Correlation Filter (with Channel and Spatial Reliability Tracking) for tracking. We calculated the speed information from the tracking information in our study. Our framework achieved 96.80% speed accuracy compared to manual observation of speeds. / Includes bibliography. / Dissertation (Ph.D.)--Florida Atlantic University, 2019. / FAU Electronic Theses and Dissertations Collection

For the last several years, convolutional neural network (CNN) based object detection systems have used a regression technique to predict improved object bounding boxes based on an initial proposal using low-level image features extracted from the CNN. In spite of its prevalence, there is little critical analysis of bounding-box regression or in-depth performance evaluation. This thesis surveys an array of techniques and parameter settings in order to further optimize bounding-box regression and provide guidance for its implementation. I refute a claim regarding training procedure, and demonstrate the effectiveness of using principal component analysis to handle unwieldy numbers of features produced by very deep CNNs.

In pattern recognition applications, it is usually important that the same identification be given for a pattern, independent of a variety of positions, rotations and /or distortions of the pattern within the recognition device's field of view. This research relates to development of a preprocessor for a neural network character recognition system, where the role of the preprocessor is to assist in minimizing the difficulties related to variations of position and rotations of a character within the field of view. The preprocessor explored here was suggested in 1970' (Lendaris & Stanly, 1970), and is implemented here with more recent advances in neural network and discrete computation technologies. The preprocessor consists of calculating the two-dimensional Fourier transform of the image (current hardware technology allows this to occur in less than 100 ms for a 256x256 pixels image , on a PC based machine with accelerator card), and then taking certain measurements on the transformed image. These measurements are given to the neural network, which processes the data to provide the character identification. Introduction of the preprocessor is shown to yield a great reduction in sensitivity to image translation and/or rotation.