Thesis (M. Phil.)--University of Hong Kong, 2005. / Title proper from title frame. Also available in printed format.

Thesis (doctoral)--University of Tartu, 2005. / Vita. Includes bibliographical references.

No description available.

Made available in DSpace on 2014-06-11T19:22:31Z (GMT). No. of bitstreams: 0 Previous issue date: 2011-12-07Bitstream added on 2014-06-13T19:28:02Z : No. of bitstreams: 1 tiago_mm_me_ilha.pdf: 962669 bytes, checksum: 4988399c15f758626b264c1adb577b2f (MD5) / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / Um dos setores de grande importância na indústria frigorífica é o responsável pelo esquarte- jamento de aves, no qual peças inteiras são separadas em partes menores para comercialização. O processo de esquartejamento pode ser feito de forma automática, através de máquinas de corte, ou por trabalhadores, que cortam as aves utilizando uma serra circular. Por ser um tra- balho manual e envolver uma lâmina de corte, a periculosidade desse tipo de trabalho é alta, de maneira que mesmo com o uso de uma luva de aço inox como equipamento de proteção, costumam ocorrer acidentes que podem variar desde pequenos cortes até amputação de parte da mão do trabalhador atingido. Neste trabalho, é apresentado um método de análise de sinais para evitar que esse tipo de acidente ocorra. Esse sistema baseia-se na análise dos sinais acústicos envolvidos gerados durante esse processo e são utilizados para desligar o motor que impulsiona a serra e acionar um sistema de frenagem em casos quando houver a ocorrência de acidentes. O problema é abordado utilizando inicialmente um filtro digital e, posteriormente, com as técni- cas de análise multirresolução apresentadas pelas wavelets. Além disso, empregou-se também a decomposição de modo empírico, que também realiza uma análise multirresolução dos sinais decompondo os mesmos em funções de modo intrínseco. Visando detectar o maior número possível de toques suaves de luva na serra sem que cortes de ossos de frango fossem confundi- dos com toques de luva, o sistema apresentou um índice de acertos de aproximadamente 70%, havendo a ocorrência de apenas 2% de falsos positivos. Além desse problema, abordou-se o caso de detecção de trinca em ovos, no qual o objetivo era separar ovos trincados de ovos in- teiros utilizando um sistema barato e eficiente... / One of the most important sectors in the meatpacking industry is chicken quartering, where whole pieces are cut into smaller ones. The quartering process can be done by automatic ma- chines or by manually cutting the chickens using a circular saw. The manual technique imposes physical risks for the workers, which wear protective stainless steel gloves. Small injuries or, in the worst case, amputation of part of the hand can occur in the event of an accident. In this work, we propose a methodology to prevent this type of accident, which is based on the anal- ysis of the acoustic signals generated during this process. In the event of an accident, the saw touches the metal glove, the acoustic signals are processed and used to turn off the engine that drives the saw and trigger a braking system. The problem is firstly analyzed using a digital filter and then with multiresolution techniques by wavelet analysis. In addition, the empirical mode decomposition technique is also employed, which also performs multiresolution analysis of sig- nals. These three techniques are implemented and compared. The method presented a 70% of successful detection of light touches of saw/glove and 2% of false positives, when a normal cut operation is detected as a saw/glove touch, in general occurring when cutting specific parts of bone. Besides this problem, the case of eggshell crack detection is studied, where the goal was to separate cracked eggs from intact eggs using an inexpensive and efficient system. A solenoid was used as a source of mechanical excitation and the resulting acoustic signals were acquired and processed. The same signal processing techniques were employed and compared, with small changes in parameters. As a result, it was possible to detect 80% of cracked eggs and 100% of intact eggs. The multiresolution technique... (Complete abstract click electronic access below)

Master of Science / Department of Computer Science / William H. Hsu / This thesis addresses the tasks of detecting vegetation and classifying plants into target crops and weeds using combinations of machine learning and pattern recognition algorithms and models. Solutions to these problems have many useful applications in precision agriculture, such as estimating the yield of a target crop or identifying weeds to help automate the selective application of weedicides and thereby reducing cost and pollution. The novel contribution of this work includes development and application of image processing and computer vision techniques to create training data with minimal human intervention, thus saving substantial human time and effort. All of the data used in this work was collected from corn fields and is in the RGB format. As part of this thesis, I first discuss several steps that are part of a general methodology and data science pipeline for these tasks, such as: vegetation detection, feature engineering, crop row detection, training data generation, training, and testing. Next, I develop software components for segmentation and classification subtasks based on extant image processing and machine learning algorithms. I then present a comparison of different classifier models developed through this process using their Receiver Operating Characteristic (ROC) curves. The difference in models lies in the way they are trained - locally or globally. I also investigate the effect of the altitude at which data is collected on the performance of classifiers. Scikit-learn, a Python library for machine learning, is used to train decision trees and other classification learning models. Finally, I compare the precision, recall, and accuracy attained by segmenting (recognizing the boundary of) plants using the excess green index (ExG) with that of a learned Gaussian mixture model. I performed all image processing tasks using OpenCV, an open source computer vision library.

Namibia, Juttadinteria and Dracophilus, the three genera belonging to the subtribe Dracophilinae Schwantes (Aizoaceae: Ruschioideae) were revised. Macro-morphology, leaf anatomy and micromorphology of leaf epidermides, pollen, tapetal orbicules and seed were studied in order to test taxon limits, to determine relationships between taxa, and to improve knowledge of their characteristics as well as the taxonomy of the subtribe. The investigation was based on herbarium specimens, field observations and cultivated, living plants. Phenetic cluster analyses were used to confirm species while intergeneric and interspecific relationships of the taxa so indicated were later tested by means of a phylogenetic analysis. The existence of the three genera was provisionally upheld by this study although phylogenetic analysis showed Namibia nested as a strongly supported monophyletic group within a poorly supported luttadinteria. Further work is needed to clarify whether Namibia should be sunk into Juttadinteria. The latest treatments of Juttadinteria and Dracophilus by Hartmann (2001) were supprted. However, in contrast to her latest treatment of Namibia, N. pomonae was sunk into N. cinerea and N. ponderosa was reinstated and typified. Juttadinteria was found to be a poorly resolved genus with many intergrading and overlapping characters, possible due to recent speciation. Possible subspecific groups within J. deserticola and J. simpsonii were indicated by the phenetic study but further work is needed before any formal infraspecific taxonomic rank can be assigned to them. Detailed descriptions of the nine elucidated species and the three genera as well as new keys for their identification are provided in order to facilitate further work in this group. Distribution maps are provided for the genera and species, and the subtribe. Variation of character states within the subtribe, distribution area, ecology and biology are discussed.

M.A. (African Languages) / Please refer to full text to view abstract

Nowadays, when a keyword is provided, a search engine can return a large number of web pages, which makes it difficult for people to find the right information. Web page classification is a technology that can help us to make a relevant and quick selection of information that we are looking for. Moreover, web page classification is important for companies that provide marketing and analytics platforms, because it can help them to build a healthy mix of listings on search engines and large directories. This will provide more insight into the distribution of the types of web pages their local business listings are found on, and finally will help marketers to make better-informed decisions about marketing campaigns and strategies. In this thesis we perform a literature review that introduces web page classification, feature selection and feature extraction. The literature review also includes a comparison of three commonly used classification algorithms and a description of metrics for performance evaluation. The findings in the literature enable us to extend existing classification techniques, methods and algorithms to address a new web page classification problem faced by our industrial partner SweetIQ (a company that provides location-based marketing services and an analytics platform). We develop a classification method based on SweetIQ's data and business needs. Our method includes typical feature selection and feature extraction methods, but the features we use in this thesis are largely different from traditional ones used in the literature. We test selected features and find that the text extracted from the title and snippet of a web page can help a classifier to achieve good performance. Our classification method does not require the full content of a web page. Thus, it is fast and saves a lot of space.

Variable screening techniques are fast and crude techniques to scan high-dimensional data and conduct dimension reduction before a refined variable selection method is applied. Its marginal analysis feature makes the method computationally feasible for ultra-high dimensional problems. However, most existing screening methods for classification problems are designed only for binary classification problems. There is lack of a comprehensive study on variable screening for multi-class classification problems. This research aims to fill the gap by developing variable screening for multi-class problems, to meet the need of high dimensional classification. The work has useful applications in cancer study, medicine, engineering and biology. In this research, we propose and investigate new and effective screening methods for multi-class classification problems. We consider two types of screening methods. The first one conducts screening for multiple binary classification problems separately and then aggregates the selected variables. The second one conducts screening for multi-class classification problems directly. In particular, for each method we investigate important issues such as choices of classification algorithms, variable ranking, and model size determination. We implement various selection criteria and compare their performance. We conduct extensive simulation studies to evaluate and compare the proposed screening methods with existing ones, which show that the new methods are promising. Furthermore, we apply the proposed methods to four cancer studies. R code has been developed for each method.

This thesis is a preliminary attempt to consolidate materials pertaining to Lau fish taxonomy (North Malaita, Solomon Islands). Data utilized come from two sources: those collected by Maranda and Maranda (1966-1968) and those collected by the author during a two-month field period (October-December 1975). Two approaches to the analysis of terminological systems are explored first. A general description of the Lau Taxonomic Universe follows in which the major components are indicated. The focus then shifts to a more detailed discussion of Lau Fish taxonomy. Material presented here takes three forms: (1) A comprehensive list of fish identified according to biological classifications. (2) A summary of data obtained from informants' Memory Lists of fish names. The problems of taxonomic inclusion and equivalence are considered. (3) A discussion of those data traditionally regarded as "Non-Taxonomic Terminology". Distinctive Features are then considered and some examples given. Upper Level Taxa are discussed first. Following this, the Features and Criteria for the classification of Lower Level Taxa are outlined. Suggestions for further inquiry and propositions concerning analytical avenues constitute the final portion of this presentation. / Arts, Faculty of / Anthropology, Department of / Graduate