Mapping land use in a large agricultural basin : a comparison between classification techniques

In order to facilitate the better management of river basin resources, the Glenelg-Hopkins region in south-east Australia required an accurate and up to date land use map. Land use has a major impact on Australia's natural resources including its soil, water, flora and fauna and plays a major role in determining basin health. Inappropriate land use and practices have contributed to extensive dryland salinity and water quality problems. Land use data is often required for environmental models and in most cases the reliability of model outputs is dependent on the spatial detail and accuracy of the land use mapping. This paper examines methods to obtain an up to date land use map and a detailed accuracy assessment using Landsat ETM+ data for a regional basin. A multi-source based approach allowed the collection of 4817 ground truth data points from the field investigation. This enabled researchers to (i) incorporate a full range of information into digital image analysis with significant improvements in accuracy and (ii) hold sufficient independent references for an accurate error assessment. Classification accuracy was significantly improved using a stratification design, in which the region is sub-divided into smaller homogenous areas as opposed to a full scene classification technique. The overall classification accuracy was 84% (KHAT= 0.833) for the stratified approach compared to 76% (KHAT= 0.743) for the full scene classification. Effective assessment, planning and management of basins are dependent on a sound knowledge of the distribution and variability of land use.

Do we need a new classification system for female sexual dysfunction? A comment on the 1999 consensus classification system.

This Article does not have an abstract.

A preliminary study of 16S rRNA sequence variation in Australia Macrobrachium shrimps (Palaemonidae: Decapoda) reveals inconsistencies in their current classification

The systematic relationships among Australian palaemonid shrimps have been the subject of speculation for some time. A preliminary phylogenetic study was undertaken to clarify the relationships of five species, Macrobrachium intermedium (Stimpson), M. australiense (Holthuis), M. atactum (Riek), M. rosenbergii (de Man) and Palaemon serenus (Heller), using 16S rRNA mitochondrial gene sequences. Phylogenetic analyses indicated inconsistencies with the current classification in two respects. First, M. intermedium formed a very well-supported clade with P. serenus distinct from M. australiense, M. atactum and M. rosenbergii. Second, the two species from inland Australia, M. australiense and M. atactum, showed a high level of genetic similarity over a substantial geographic range, suggesting that they may represent conspecific populations. The taxonomic and biogeographic implications of these findings for Macrobrachium in Australia are discussed.

Intelligent animal fiber classification with artificial neural networks

Artificial neural networks (ANN) are increasingly used to solve many problems related to pattern recognition and object classification. In this paper, we report on a study using artificial neural networks to classify two kinds of animal fibers: merino and mohair. We have developed two different models, one extracting nine scale parameters with image processing, and the other using an unsupervised artificial neural network to extract features automatically, which are determined in accordance with the complexity of the scale structure and the accuracy of the model. Although the first model can achieve higher accuracy, it requires more effort for image processing and more prior knowledge, since the accuracy of the ANN largely depends on the parameters selected. The second model is more robust than the first, since only raw images are used. Because only ordinary optical images taken with a microscope are employed, we can use the approach for many textile applications without expensive equipment such as scanning electron microscopy.

Objective classification of fabric pilling based on the two-dimensional discrete wavelet transform

A number of methods for automated objective ratings of fabric pilling based on image analysis are described in the literature. The periodic structure of fabrics makes them suitable candidates for frequency domain analysis. We propose a new method of frequency domain analysis based on the two-dimensional discrete wavelet transform to objectively measure pilling intensity in sample images. We present a preliminary evaluation of the proposed method based on analysis of two series of standard pilling evaluation test images. The initial results suggest that the proposed method is feasible, and that the ability of the method to discriminate between levels of pilling intensity depends on the wavelet analysis scale being closely matched to the fabric interyarn pitch. We also present a heuristic method for optimal selection of an analysis wavelet and associated analysis scale.

A new algorithm for the Petrov classification of the Weyl tensor

A new algorithm for the Petrov classification of the Weyl tensor is introduced. It is similar to the Letniowski-McLenaghan algorithm [1] when someof the ¥'s are zero, but offers a completely new approach when all of the ¥'s are nonzero. In all cases, new code in Maple has been implemented.

A new algorithm for the Segre classification of the trace-free Ricci tensor

A new algorithm, based on the introduction of new spinor quantities, for the Segre classification of the trace-free Ricci tensor is presented. It is capable of automatically distinguishing between the two Segre types [1,1(11)] and [(1,1)11] where all other known algorithms fail to do so.

Evaluating the robustness of objective pilling classification with the two-dimensional discrete wavelet transform

Previously, we proposed a new method of frequency domain analysis based on the two-dimensional discrete wavelet transform to objectively measure pilling intensity in sample fabric images. We have further evaluated this method, and our results indicate that it is robust to small horizontal and/or vertical translations and to significant variations in the brightness of the image under analysis, and is sensitive to rotation and to dilation of the image. These results suggest that as long as precautions are taken to ensure fabric test samples are imaged under consistent conditions of weave/knit pattern alignment (rotation) and apparent interyarn pitch (dilation), the method will yield repeatable results.

The classification of lands managed for conservation : existing and proposed frameworks, with particular reference to Australia

Comprehensive classification systems to accurately account for lands managed for biodiversity conservation, are an essential component of conservation planning and policy. The current international classification systems for lands managed for nature conservation are reviewed, with a particular emphasis on Australia. The need for a broader, all-encompassing, categorisation of lands managed for conservation is presented and a proposed broader categorisation system is developed—the Conservation Lands Classification. This classification system has the advantage of incorporating data on both tenure and protection mechanisms and has been applied in this paper using conservation lands in three Australian jurisdictions as examples. It is envisaged that this method of classification has the potential to significantly improve the ability to measure current and future trends in nature conservation across all land types at a variety of scales and hence is put forward in order to stimulate discussion on this important topic.

Hydrology or floristics? Mapping and classification of wetlands in Victoria, Australia, and implications for conservation planning

A national approach to the conservation of biodiversity in Australia’s freshwater ecosystems is a high priority. This requires a consistent and comprehensive system for the classification, inventory, and assessment of wetland ecosystems. This paper, using the State of Victoria as a case study, compares two classification systems that are commonly utilized to delineate and map wetlands—one based on hydrology (Victorian Wetland Database [VWD]) and one based on indigenous vegetation types and other natural features (Ecological Vegetation Classes [EVC]). We evaluated the extent of EVC mapping of wetlands relative to the VWD classification system using a number of datasets within a geographical information system. There were significant differences in the coverage of extant EVCs across bioregions, different-sized wetlands, and VWD wetland types. Resultant depletion levels were markedly different when examined using the two systems, with depletion levels, and therefore perceived conservation status, of EVCs being significantly higher. Although there is little doubt that many wetland ecosystems in Victoria are in fact threatened, the extent of this threat cannot accurately be determined by relying on the EVC mapping as it currently stands. The study highlighted the significant impact wetland classification methods have in determining the conservation status of freshwater ecosystems.

Reconsidering the benefits of equitable classification

A deductive system that enables us to derive many legal rules from a few principles makes the law more, rather than less certain, since this approach parallels the actual process by which judicial decisions are reached. Uncertainty as to the meaning of equity in the law is inevitably . .. due to the absence of legal guidance for the standard of moral values to be observed in transactions . .. 1

Characterization and application of objective pilling classification to patterned fabrics

Previously, the authors proposed a new, simple method of frequency domain analysis based on the two-dimensional discrete wavelet transform to objectively measure the pilling intensity in sample fabric images. The method was further characterized, and the results obtained indicate that standard deviation and variance are the most appropriate measures of the dispersion of wavelet details coefficients for analysis, that the relationship between wavelet analysis scale and fabric inter-yarn pitch was empirically confirmed, and, that fabrics with random patterns do not appear to impact on the effectiveness of the analysis method.

A novel field learning algorithm for dual imbalance text classification

Fish-net algorithm is a novel field learning algorithm which derives classification rules by looking at the range of values of each attribute instead of the individual point values. In this paper, we present a Feature Selection Fish-net learning algorithm to solve the Dual Imbalance problem on text classification. Dual imbalance includes the instance imbalance and feature imbalance. The instance imbalance is caused by the unevenly distributed classes and feature imbalance is due to the different document length. The proposed approach consists of two phases: (1) select a feature subset which consists of the features that are more supportive to difficult minority class; (2) construct classification rules based on the original Fish-net algorithm. Our experimental results on Reuters21578 show that the proposed approach achieves better balanced accuracy rate on both majority and minority class than Naive Bayes MultiNomial and SVM.

Parameter estimation of one-class SVM on imbalance text classification

Compared with conventional two-class learning schemes, one-class classification simply uses a single class for training purposes. Applying one-class classification to the minorities in an imbalanced data has been shown to achieve better performance than the two-class one. In this paper, in order to make the best use of all the available information during the learning procedure, we propose a general framework which first uses the minority class for training in the one-class classification stage; and then uses both minority and majority class for estimating the generalization performance of the constructed classifier. Based upon this generalization performance measurement, parameter search algorithm selects the best parameter settings for this classifier. Experiments on UCI and Reuters text data show that one-class SVM embedded in this framework achieves much better performance than the standard one-class SVM alone and other learning schemes, such as one-class Naive Bayes, one-class nearest neighbour and neural network.

The need for low bias algorithms in classification learning from large data sets

This paper reviews the appropriateness for application to large data sets of standard machine learning algorithms, which were mainly developed in the context of small data sets. Sampling and parallelisation have proved useful means for reducing computation time when learning from large data sets. However, such methods assume that algorithms that were designed for use with what are now considered small data sets are also fundamentally suitable for large data sets. It is plausible that optimal learning from large data sets requires a different type of algorithm to optimal learning from small data sets. This paper investigates one respect in which data set size may affect the requirements of a learning algorithm — the bias plus variance decomposition of classification error. Experiments show that learning from large data sets may be more effective when using an algorithm that places greater emphasis on bias management, rather than variance management.