Steel industry is a highly polluted industry and one of the most important stationary sources in Kaohsiung City. The steel manufacturing process could emit a huge amount of particles, such as the sintering process, the blast furnace operation, and the raw material handling process. Suspended particles emitted from steel industry could deteriorate ambient air quality and cause adverse effects on human health. In order to understand the impact of steel industry on ambient air quality in Siaogang District and to identify potential pollution sources, this study selected a integrated steel manufacturing plant located at Siaogang District to conduct a sampling protocol of suspended particulate matter (PM) at ambient sites (A1~A5) and sensitive sites (S1~S5) from July 2011 to March 2012. The size distribution of suspended particles in four seasons was measured with PM10 high-volume samplers, dichotomous samplers, and MOUDI for 3 days (24 hours for single sampling), and dustfall samplers for one month, to investigate the spatial distribution and temporal variation of PM concentration. After sampling, the physicochemical properties of PM, including mass concentration, particle size distribution, dustfall concentration, water-soluble ionic species, metallic elements, and carbonaceous contents, were further analyzed. Field measurement of ambient PM showed that the averaged ambient PM10 concentration (53.54 - 203.56 £gg/m3) were higher than sensitive sites (55.06 - 140.07 £gg/m3) and the averaged ambient PM2.5 concentration of ambient (23.10 - 120.21£gg/m3) were higher than sensitive sites (12.52 - 65.62 £gg/m3). No matter ambient or sensitive sites, it showed a tendency of lower concentration in summer, indicating that concentration variation of PM10 and PM2.5 were highly affected by meteorological factors (such as wind direction, wind speed, and relative humidity) in Siaogang District. Furthermore, a t-test result showed that ambient and sensitive sites have similar pollution sources since the p-values were in significantly different. Chemical analysis of PM results showed that the most abundant water-soluble ionic species of PM at the ambient and sensitive sites were secondary inorganic aerosols (SO42-, NO3-, and NH4+) and [NO3-]/[SO42-] showed that ionic species were mainly emitted from stationary sources. Fe, Al, K and Ca were the major metallic elements of this study, and the major pollution sources contain industries, traffics, and road dusts. Additionally, the raw material handling process was the major pollution source of PM. Correlation analysis of OC and EC showed that PM at ambient and sensitive sites were originated from primary sources, such as vehicles, industries, road dusts, and human activities. Results obtained from PCA and CMB receptor modeling showed that both PM2.5 and PM10 highly correlated with wind direction in different season and the major pollution sources were industry pollution (including petroleum refineries, power plants, waste incinerators, consistent operating steel mills and electric arc furnace steel mills, etc.), followed by local traffics and derivative. Furthermore, marine aerosols were one of the important pollution sources at sensitive sites (S1, S4, and S5) where close to the ocean.

Financial time-series forecasting has long been discussed because of its vitality for making informed investment decisions. This kind of problem, however, is intrinsically challenging due to the data dynamics in nature. Most of the research works in the past focus on artificial neural network (ANN)-based approaches. It has been pointed out that such approaches suffer from explanatory power and generalized prediction ability though. The objective of this research is thus to propose a hybrid approach for stock price forecasting. Independent component analysis (ICA) is employed to reveal the latent structure of the observed time-series and remove noise and redundancy in the structure. It further assists clustering analysis. Support vector regression (SVR) models are then applied to enhance the generalization ability with separate models built based on the time-series data of companies in each individual cluster. Two experiments are conducted accordingly. The results show that SVR has robust accuracy performance. More importantly, SVR models with ICA-based clustered data perform better than the single SVR model with all data involved. Our proposed approach does enhance the generalization ability of the forecasting models, which justifies the feasibility of its applications.

Blocking artifact, characterized by visually noticeable changes in pixel values along block boundaries, is a common problem in block-based image/video compression, especially at low bitrate coding. Various post-processing techniques have been proposed to reduce blocking artifacts, but they usually introduce excessive blurring or ringing effects. This paper proposes a self-learning-based image/ video deblocking framework via properly formulating deblocking as an MCA (morphological component analysis)-based image decomposition problem via sparse representation. The proposed method first decomposes an image/video frame into the low-frequency and high-frequency parts by applying BM3D (block-matching and 3D filtering) algorithm. The high-frequency part is then decomposed into a ¡§blocking component¡¨ and a ¡§non-blocking component¡¨ by performing dictionary learning and sparse coding based on MCA. As a result, the blocking component can be removed from the image/video frame successfully while preserving most original image/video details. Experimental results demonstrate the efficacy of the proposed algorithm.

Laser based underwater triangulation ranging is sensitive to the environmental conditions and laser beam profile. Also, its ranging quality is greatly affected by the algorithm choices for peak detection and for image processing. By utilizing the merging least-squares approximation for laser image processing, it indeed succeeds in increasing quality of triangulation ranging in water; however, this result was obtained on the use of a laser beam with nearly circular cross-section. Therefore, by using an ellipse-like laser beam cross-section for range finding, we are really interested in understanding the quality of range finding with different peak detection algorithms. Besides, the ellipse orientation of the laser spot projected on the image plane would be various. We are also interested in learning about the relationship between the ellipse orientation and the quality of range finding. In this study, peak detection algorithms are investigated by considering four different laser beam cross-sections which are ircle, horizontal ellipse, oblique ellipse, and vertical ellipse. First, we employ polynomial regression for processing laser image to study the effect of polynomial degree on quality of triangulation ranging. It was found that the linear regression achieves the best ranging quality than others. Then, according to this result, the ranging quality associated with peak detection is evaluated by employing three different algorithms which are the illumination center, twice illumination center and the illumination center with principal component analysis. We found that the ranging quality by using the illumination center with principal component analysis is the best, next is twice illumination center, and last the illumination center. This result indicates that the orientation of elliptical laser beam has an influential effect on the quality of range finding. In addition, the ranging quality difference among peak detection algorithms is significantly reduced by implementing the merging least-squares approximation rlaser image processing. This result illustrates that the merging least-squares approximation does reduce the effect of peak detection algorithm on the quality of range finding.

For many undersea research application scenarios, instruments need to be deployed for more than one month which is the basic time interval for many phenomena. With limited power supply and memory, management strategies are crucial for the success of data collection. For acoustic recording of undersea activities, in general,either preprogrammed duty cycle is configured to log partial time series,or spectrogram of signal is derived and stored,to utilize the available memory storage efficiently.To overcome this limitation, we come up with an algorithm to classify different and store only the sound data of interest. Features like characteristic frequencies, large amplitude of selected frequencies or intensity threshold are used to identify or classify different patterns. On main limitation for this type of approaches is that the algorithm is generally range-dependent, as a result, also sound-level-dependent. This type of algorithms will be less robust to the change of the environment.One the other hand, one interesting observation is that when human beings look at the spectrogram, they will immediately tell the difference between two patterns. Even though no knowledge about the nature of the source, human beings still can discern the tiny dissimilarity and group them accordingly. This suggests that the recognition and classification can be done in spectrogram as a recognition problem. In this work, we propose to modify Principal Component Analysis by generating feature points from moment invariant and sound Level variance, to classify sounds of interest in the ocean. Among all different sound sources in the ocean, we focus on three categories of our interest, i.e., rain, ship and whale and dolphin. The sound data were recorded with the Passive Acoustic Listener developed by Nystuen, Applied Physics Lab, University of Washington. Among all the data, we manually identify twenty frames for each cases, and use them as the base training set. Feed several unknown clips for classification experiments, we suggest that both point-based feature extraction are effective ways to describe whistle vocalizations and believe that this algorithm would be useful for extracting features from noisy recordings of the callings of a wide variety of species.

This thesis investigates the determinants of financial development in Turkey. Principle Component Analysis (PCA) is employed in order to examine the main determinants of financial sector development and shed light on the structure of the financial system in Turkey. The empirical studies on financial development suffer from the measurement problem. This study aims to remedy the measurement problem by providing proxies that explain different aspects of financial development more accurately than other proxies used in the extant literature. Hence, the present study constitutes a strong basis for studies that rely on measuring financial development in Turkey.

High-dimensional data take place from digital image processing, gene expression micro arrays, neuronal population activities to financial time series. Dimensionality Reduction - extracting low dimensional structure from high dimension - is a key problem in many areas like information processing, machine learning, data mining, information retrieval and pattern recognition, where we find some data reduction techniques. In this thesis we will give a survey about modern data reduction techniques, representing the state-of-the-art of theory, methods and application, by introducing the language of mathematics there. This needs a special care concerning the questions of, e.g., how to understand discrete structures as manifolds, to identify their structure, preparing the dimension reduction, and to face complexity in the algorithmically methods. A special emphasis will be paid to Principal Component Analysis, Locally Linear Embedding and Isomap Algorithms. These algorithms are studied by a research group from Vilnius, Lithuania and Zeev Volkovich, from Software Engineering Department, ORT Braude College of Engineering, Karmiel, and others. The main purpose of this study is to compare the results of the three of the algorithms. While the comparison is beeing made we will focus the results and duration.

No description available.

Functional data refer to data which consist of observed functions or curves evaluated at a finite subset of some interval. In this dissertation, we discuss statistical analysis, especially classification and regression when data are available in function forms. Due to the nature of functional data, one considers function spaces in presenting such type of data, and each functional observation is viewed as a realization generated by a random mechanism in the spaces. The classification procedure in this dissertation is based on dimension reduction techniques of the spaces. One commonly used method is Functional Principal Component Analysis (Functional PCA) in which eigen decomposition of the covariance function is employed to find the highest variability along which the data have in the function space. The reduced space of functions spanned by a few eigenfunctions are thought of as a space where most of the features of the functional data are contained. We also propose a functional regression model for scalar responses. Infinite dimensionality of the spaces for a predictor causes many problems, and one such problem is that there are infinitely many solutions. The space of the parameter function is restricted to Sobolev-Hilbert spaces and the loss function, so called, e-insensitive loss function is utilized. As a robust technique of function estimation, we present a way to find a function that has at most e deviation from the observed values and at the same time is as smooth as possible.

The enzymatic hydrolysis of cellulose to glucose by cellulases is one of the major steps in the conversion of lignocellulosic biomass to biofuel. This hydrolysis by cellulases, a heterogeneous reaction, currently suffers from some major limitations, most importantly a dramatic rate slowdown at high degrees of conversion in the case of crystalline cellulose. Various rate-limiting factors were investigated employing experimental as well as computational studies. Cellulose accessibility and the hydrolysable fraction of accessible substrate (a previously undefined and unreported quantity) were shown to decrease steadily with conversion, while cellulose reactivity, defined in terms of hydrolytic activity per amount of actively adsorbed cellulase, remained constant. Faster restart rates were observed on partially converted cellulose as compared to uninterrupted hydrolysis rates, supporting the presence of an enzyme clogging phenomenon. Cellulose crystallinity is a major substrate property affecting the rates, but its quantification has suffered from lack of consistency and accuracy. Using multivariate statistical analysis of X-ray data from cellulose, a new method to determine the degree of crystallinity was developed. Cel7A CBD is a promising target for protein engineering as cellulose pretreated with Cel7A CBDs exhibits enhanced hydrolysis rates resulting from a reduction in crystallinity. However, for Cel7A CBD, a high throughput assay is unlikely to be developed. In the absence of a high throughput assay (required for directed evolution) and extensive knowledge of the role of specific protein residues (required for rational protein design), the mutations need to be picked wisely, to avoid the generation of inactive variants. To tackle this issue, a method utilizing the underlying patterns in the sequences of a protein family has been developed.