Recent research addresses the problem of data-stream mining to deal with applications that require processing huge amounts of data such as sensor data analysis and financial applications. Data-stream mining algorithms incorporate special provisions to meet the requirements of stream-management systems, that is stream algorithms must be online and incremental, processing each data record only once (or few times); adaptive to distribution changes; and fast enough to accommodate high arrival rates. We consider the problem of data-stream classification, introducing an online and incremental stream-classification ensemble algorithm, Streaming Random Forests, an extension of the Random Forests algorithm by Breiman, which is a standard classification algorithm. Our algorithm is designed to handle multi-class classification problems. It is able to deal with data streams having an evolving nature and a random arrival rate of training/test data records. The algorithm, in addition, automatically adjusts its parameters based on the data seen so far. Experimental results on real and synthetic data demonstrate that the algorithm gives a successful behavior. Without losing classification accuracy, our algorithm is able to handle multi-class problems for which the underlying class boundaries drift, and handle the case when blocks of training records are not big enough to build/update the classification model. / Thesis (Ph.D, Computing) -- Queen's University, 2008-07-15 16:12:33.221

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

This study extends an earlier M.Sc. research project on the narrow-leaved species of the genus Asclepias L. (30 species) to cover the entire subtribe Asclepiadinae sensu K. Schum. in southern Africa- (182 species in 23 genera). Two genera (Eustegia R. Br. and Pentarrhinum E. Mey.) are revised and then removed from this tribe. The remaining 177 species and 21 genera form the focus of this thesis, whose principal objectives are to reevaluate the taxonomic and evolutionary significance of various macro, micro and chemical characters and then use them to produce a classification that, more closely, reflects the overall similarity and phylogeny of the taxa involved. Species and genera are recircumscribed based on the wealth of data that has come to light since the subtribe was last revised by N.E. Brown (1907-1908) some 90 years ago. This process was supplemented by extensive field work, observations on pollination and reproductive biology, ecology, biogeography, conservation and ethnobotany. The majority of this thesis consists of a compilation of 17 papers, 12 of these published and most, but not all, of the remainder in preparation for press. Two of these papers form the bulk of the taxonomy. The first deals with what was the genus Asclepias in southern Africa. The genus is now believed to be confined to the Americas. The • Botswana, Lesotho, Namibia, South Africa and Swaziland. southern African species have diverse origins and are partitioned into 7 genera, one of them (Gomphocarpus R. Br.) resurrected, two of them (Aidomene Stopp and Aspidonepsis Nicholas & Goyder) expanded and four of them (Paulforstera, Sigridia, Bruynsia and Pachyacris) described as new. Gomphocarpus is divided into two subgenera and Aidomene into four subgenera. Three new species are also described. The second paper investigates the bulk of most of the remaining genera. Kanahia R. Br., Cordylogyne E. Mey. and Fanninia Harv. remain as is. Xysmalobium R. Br., previously a genus of 19 species in southern Africa, is reduced to three species in two subgenera. Trichocodon is segregated off from Pachycarpus E. Mey. as a new genus. While two species, previously placed in Xysmalobium, are added to Pachycarpus, but placed in the new subgenus Parapodiopsis. Parapodium E. Mey. is reduced from three to two species and Periglossum Decne. is reduced from five to three species, one of them newly described. Woodia Schltr. and Stenostelma Schltr. are both considerably expanded, mainly with species previously housed in Xysmalobium), and the former divided into two subgenera. The third paper briefly looks at the Schizoglossum E. Mey., Miraglossum Kupicha and Aspidoglossum E. Mey. Some changes are suggested but, as further work is needed, none are formalised. As a corollary to the taxonomy, secondary metabolite profiles of 38 species and 17 genera were done using Thin layer Chromatography. The results sometimes confirmed morphological patterns and sometimes were at odds with them. A trend from simple profiles to more complex profiles seems to echo the suspected phylogeny of the genera within this sub tribe. Some species and genera have greater chemical diversity than others and secondary metabolites are shown to vary considerably in different parts of a single plant. As a supplement to the above work or because they are cited elsewhere in the dissertation, published papers dealing with floral structure, the asclepiadaceous work of Rudolf Schlechter, as well as miscellaneous works in the tribe Stapelieae, are also given. / Thesis (Ph.D.)-University of Durban-Westville, 1999.

This study comprises an examination of over 4000 male specimens, including nearly all type material, a detailed study of the genitalia, and a key to the majority of the species of Tricholabiodes Radoszkowski. Thirty species and subspecies are redescribed and 22 new species described. The 22 newly described species are: T. acer, T. alveolus, T. brothersi, T. concavus, T. convexus, T. denticidatus, T. disgregus, T. femoralis, T. ferrugineus, T. indistinctus, T. inornatus, T. longicarinatus, T. liiridus, T. parallel™, T. paulocellatiis, T. petiolatus, T. protitberans, T. recurvatus, T. sinuatus, T. thisboides, T. tortilis and T. trochantalis. Tricholabiodes semisthataeformis Bischoff and T. pathzii Invrea are synonyms of T. stigmaticus Bischoff and T. pallidicornis Bischoff, respectively. Phenograms and principal component plots were derived to clarify species status, make decisions on species limits and used to determine the morphological similarity between the species. The phenetic analysis was used only as a tool, and not a final product. For the determination of species limits, which included an analysis of 447 specimens, the continuous quantitative and coded characters were analysed separately. Forty-three continuous quantitative characters were analysed either as standardized measurements (against mesosoma length) or as ratios (32), since it was not possible, even via gap coding, to code these characters. Scatterplots and a phenogram from the principal components and cluster analyses respectively, are presented. Size and shape were not particularly helpful characters in determining species limits. One hundred and twenty five coded characters were analysed in a cluster analysis and part of the final phenogram is presented. For the determination of morphological similarity between the species, a hypothetical specimen, typical of each species, was derived. Again, one hundred and twenty-five coded characters were analysed in a cluster analysis and the final phenogram is presented. Representatives from each of the species and subspecies were examined with respect to 93 coded characters. The character states were polarised using the outgroup Dasylabroides Andre. Where Tricholabiodes had all states occurring in Dasylabroides, and the primitive state could not be identified, these characters, and their states, were considered for the entire tribe, and the sister tribe of Dasylabrini, Sphaeropthalmini, was taken as the outgroup. The cladograms were constructed with the software Hennig86. The most variable characters were eliminated from the analysis. Selection of the cladogram representing the most likely phylogeny of the genus was based on parsimony, resolution of the tree, character placement on the tree, comparison of the tree with weighted/unweighted consensus trees and biogeography. The phylogeny presented, which is to be regarded only as a hypothesis, suggests that Tricholabiodes underwent nine separate radiations. The southern African species are divided into two lineages: the first divergence stems from the base of the tree while the more recent lineage stems from the apex. Evidence suggests that the genus arose in central Africa, spreading south (twice) into southern Africa, north into North Africa, west across central North Africa and east into southeastern Asia. The study has also shown that the majority of the species are restricted in their distribution, with none of Palaearctic species occurring in southern Africa, and vice versa. It is hypothesised that the present distribution of the genus is partially restricted by dispersal ability and climate. / Thesis (Ph.D.)-University of Natal, Pietermaritzburg, 1998.

Members of the genus Ledebouria Roth (Hyacinthaceae), which occur in South Africa, are revised. This genus occurs throughout Africa, India and Madagascar. 33 Species are recognized and placed into nine provisional infrageneric groups. A multidisciplinary approach was used in an attempt to provide natural groupings. The following characters were analysed; morphology, micromorphology, palynology and caryology. Aspects of ovary structure and leaf micromorphology proved especially useful in the synthesis of infrageneric and specific concepts. Keys, descriptions, illustrations, distributional, ecological, medicinal and toxicological data are provided. This study is based on plants in their natural habitat, cultivated specimens and representative herbarium specimens from herbaria in South Africa and in Europe. / Thesis (M.Sc.)-University of Natal, Pietermaritzburg, 1993.