Description:
The relatively new field of stream mining has necessitated the development of robust drift-aware algorithms that provide accurate, real time, data handling capabilities. Tools are needed to assess and diagnose important trends and investigate drift evolution parameters. In this paper, we present two new and novel visualisation techniques, Pixie and Luna graphs, which incorporate salient group statistics coupled with intuitive visual representations of multidimensional groupings over time. Through the novel representations presented here, spatial interactions between temporal divisions can be diagnosed and overall distribution patterns identified. It provides a means of evaluating in non-constrained capacity, commonly constrained evolutionary problems.
Description:
The relatively new field of stream mining has necessitated the development of robust drift-aware algorithms that provide accurate, real time, data handling capabilities. Tools are needed to assess and diagnose important trends and investigate drift evolution parameters. In this paper, we present two new and novel visualisation techniques, Pixie and Luna graphs, which incorporate salient group statistics coupled with intuitive visual representations of multidimensional groupings over time. Through the novel representations presented here, spatial interactions between temporal divisions can be diagnosed and overall distribution patterns identified. It provides a means of evaluating in non-constrained capacity, commonly constrained evolutionary problems.
Description:
In this paper we propose a novel technique of using a hybrid evolutionary method, which uses a combination of genetic algorithm and matrix based solution methods such as QR factorization. The training of the model is based on a layer based hierarchical structure for the architecture and the weights for the Artificial Neural Network classifier. The architecture for the classifier is found using a binary search type procedure. The hierarchical structured algorithm (EALS-BT) is also a hybrid, because it combines the Genetic Algorithm based method with the Matrix based solution method for finding weights. A heuristic segmentation algorithm is initially used to over segment each word. Then the segmentation points are passed through the rule-based module to discard the incorrect segmentation points and include any missing segmentation points. Following the segmentation the contour is extracted between two correct segmentation points. The contour is passed through the feature extraction module that extracts the angular features, after which the EALS-BT algorithm finds the architecture and the weights for the classifier network. These recognized characters are grouped into words and passed to a variable length lexicon that retrieves words that have the highest confidence value.
Description:
In this paper, we propose and describe sProxy - smart proxy, a software tool in Web Services transaction. sProxy acts as a gateway between transaction management systems and Web services which implements a key abstraction of proxy management systems. This enables to perform transactions in the loosely coupled environment i.e. loose coupling among services. Proxies are useful to invoke Web services to allow an easy programming model that facilitates the serialization and transmission of service invocations. Our proposed model supports relaxation of traditional ACID properties with existing commit and recovery protocols. The model works on non-ACID type of transactions which encapsulates Web services. It also uses multithreading proxies to check and update transaction simultaneously. The proposed model solves the current problems with distributed computational activities which involves both transactions and Web Services. The proposed model is more abstract and generic as demonstrated in the paper.
Description:
In this paper, we propose and describe sProxy - smart proxy, a software tool in Web Services transaction. sProxy acts as a gateway between transaction management systems and Web services which implements a key abstraction of proxy management systems. This enables to perform transactions in the loosely coupled environment i.e. loose coupling among services. Proxies are useful to invoke Web services to allow an easy programming model that facilitates the serialization and transmission of service invocations. Our proposed model supports relaxation of traditional ACID properties with existing commit and recovery protocols. The model works on non-ACID type of transactions which encapsulates Web services. It also uses multithreading proxies to check and update transaction simultaneously. The proposed model solves the current problems with distributed computational activities which involves both transactions and Web Services. The proposed model is more abstract and generic as demonstrated in the paper.
Description:
This article describes an argumentation shell to support the formulation, representation and drafting of legal arguments. The shell can be used to capture generic arguments in many legal domains as well as to assist decision-makers in constructing their own actual arguments . The shell demonstrates that knowledge represented using the generic/actual argument model (GAAM) (a variant of Toulmin's argument structure) can be used to: (a) support the development of complex arguments, (b) add context and increase specificity for the retrieval of relevant documents, (c) incorporate background knowledge, (d) assist in the drafting of documents that represent arguments made, and (e) provide a structure for complex inferences requiring a range of mechanisms. The shell can be used to support decision making in a range of legal domains, including discretionary domains.
Description:
Determining who has access to personal data is an ongoing problem facing information system entities. The establishment of trust and its representation for known and unknown entities within the system further complicates access control rights allocation. One unique solution is through the application of graph representation to aid in the identification and management of privacy, trust and security requirements. Graphs provide a much better mental map than would textual information. In this paper we use graphs to represent informational relations concerning trust levels between entities for privacy and security requirements.
Description:
We deal with the problem of labeling the vertices, edges and faces of a special class of plane graphs with 3-sided internal faces in such a way that the label of a face and the labels of the vertices and edges surrounding that face all together add up to the weight of that face. These face weights then form an arithmetic progression with common difference d.
Description:
This paper presents a computational approach for predicting the Australian stock market index AORD using multi-layer feed-forward neural networks front the time series data of AORD and various interrelated markets. This effort aims to discover an effective neural network, or a set of adaptive neural networks for this prediction purpose, which can exploit or model various dynamical swings and inter-market influences discovered from professional technical analysis and quantitative analysis. Within a limited range defined by our empirical knowledge, three aspects of effectiveness on data selection are considered: effective inputs from the target market (AORD) itself, a sufficient set of interrelated markets,. and effective inputs from the interrelated markets. Two traditional dimensions of the neural network architecture are also considered: the optimal number of hidden layers, and the optimal number of hidden neurons for each hidden layer. Three important results were obtained: A 6-day cycle was discovered in the Australian stock market during the studied period; the time signature used as additional inputs provides useful information; and a basic neural network using six daily returns of AORD and one daily, returns of SP500 plus the day of the week as inputs exhibits up to 80% directional prediction correctness.
Description:
This paper presents a computational approach for predicting the Australian stock market index AORD using multi-layer feed-forward neural networks front the time series data of AORD and various interrelated markets. This effort aims to discover an effective neural network, or a set of adaptive neural networks for this prediction purpose, which can exploit or model various dynamical swings and inter-market influences discovered from professional technical analysis and quantitative analysis. Within a limited range defined by our empirical knowledge, three aspects of effectiveness on data selection are considered: effective inputs from the target market (AORD) itself, a sufficient set of interrelated markets,. and effective inputs from the interrelated markets. Two traditional dimensions of the neural network architecture are also considered: the optimal number of hidden layers, and the optimal number of hidden neurons for each hidden layer. Three important results were obtained: A 6-day cycle was discovered in the Australian stock market during the studied period; the time signature used as additional inputs provides useful information; and a basic neural network using six daily returns of AORD and one daily, returns of SP500 plus the day of the week as inputs exhibits up to 80% directional prediction correctness.
Description:
In this paper, we will present a new approach of using link information to improve the accuracy and efficiency of web classification. However, different from others, we only use the mappings between linked documents and their own class or classes. In this case, we only need to add a few features called linked-class features into the datasets. We apply SVM and BoosTexter for classification. We show that the classification accuracy can be improved based on mixtures of ordinary word features and out-linked-class features. We analyze and discuss the reason of this improvement.
Description:
In this paper, we will present a new approach of using link information to improve the accuracy and efficiency of web classification. However, different from others, we only use the mappings between linked documents and their own class or classes. In this case, we only need to add a few features called linked-class features into the datasets. We apply SVM and BoosTexter for classification. We show that the classification accuracy can be improved based on mixtures of ordinary word features and out-linked-class features. We analyze and discuss the reason of this improvement.
Description:
The first version of this essay was jointly delivered by the authors as a colloquium lecture at the University of Ballarat on 24 November, 2004. A second, expanded and illustrated version was published in German in the Mitteilungen der Deutschen Mathematikervereinigung early in 2005. Because of the very positive feedback, the authors decided it would be useful to publish a version in English in a computing journal. The purpose of the essay is to provide advice and information to authors of articles about publishing in scholarly journals from an editor's perspective. Of particular importance are remarks about etiquette.
Description:
In this study we investigated a hybrid model based on the Discrete Gradient method and an evolutionary strategy for determining the weights in a feed forward artificial neural network. Also we discuss different variants for hybrid models using the Discrete Gradient method and an evolutionary strategy for determining the weights in a feed forward artificial neural network. The Discrete Gradient method has the advantage of being able to jump over many local minima and find very deep local minima. However, earlier research has shown that a good starting point for the discrete gradient method can improve the quality of the solution point. Evolutionary algorithms are best suited for global optimisation problems. Nevertheless they are cursed with longer training times and often unsuitable for real world application. For optimisation problems such as weight optimisation for ANNs in real world applications the dimensions are large and time complexity is critical. Hence the idea of a hybrid model can be a suitable option. In this study we propose different fusion strategies for hybrid models combining the evolutionary strategy with the discrete gradient method to obtain an optimal solution much quicker. Three different fusion strategies are discussed: a linear hybrid model, an iterative hybrid model and a restricted local search hybrid model. Comparative results on a range of standard datasets are provided for different fusion hybrid models.
Description:
A statistical database system is a system that contains information about individuals, companies or organisations that enables authorized users to retrieve aggregate statistics such as mean and count. The regulation of a statistical database involves limiting the use of the database so that no sequence of queries is sufficient to infer protected information about an individual. The database is said to be compromised when individual confidential information is obtained as a result of a statistical query. Devices to protect against compromise include adding noise to the data or restricting a query. While effective, these techniques are sometimes too strong in that legitimate compromises for reasons of public safety are always blocked. Further, a statistical database can be often be compromised with some knowledge about the database attributes (working knowledge), the real world (supplementary knowledge) or the legal system (legal knowledge). In this paper we illustrate that a knowledge based system that represents working, supplementary and legal knowledge can contribute to the regulation of a statistical database.
Description:
Real-time solution of the Discrete Element Method is a computational challenge that is hardly achievable on standard PCs, especially when a large number of triangular shaped particles are involved. This paper presents a scalable architecture, including a domain decomposition technique, of a GPU accelerator for the two-dimensional Discrete Element Method for triangular shaped particles. This approach achieved a speed up of about 140 times as a single core and about 80 after domain decomposition on a consumer level GPU compared to a similar algorithm run on a fast desktop PC.
Description:
Over 2007-2008, a pedagogy subject in a pre-service teacher education degree was (re)designed to help students develop their understandings and skills and a wider, more critical appreciation of the work of teachers and approaches to curriculum. The rationale for designing and including the online modules in the subject was to develop information and communication technology (ICT) skills, and to deliver a blended learning approach, argued by some to be more effective, that is, have more advantages than traditional approaches. In this paper, the face-to-face teaching alongside the eLearning that occurred in the blended learning approach is analysed using Tom Reeves and John Hedberg's model (2003) for evaluating interactive learning systems. Arguably, this evaluation model can be usefully applied to higher education teaching that is not fully online, and can help to comprise an integral part of an action research approach. This paper is a 'proof of concept' piece, demonstrating the applicability of the model to a blended learning course. Demonstrating the application of Reeves and Hedberg's model fills a knowledge void on the literature surrounding blended learning. [ABSTRACT FROM AUTHOR]
Description:
Predicting the secondary structure of proteins from a primary sequence alone has been variously approached from either a classification or a generative model perspective. The most prominent classification methods have used neural networks, which involves mappings from a local window of residues in the sequence to the structural state of the central residue in the window, thus capturing the local interactions effectively. However, they fail to capture distant interactions among residues. The generative models based on Bayesian segmentation capture sequence structure relationships using generalized hidden Markov models with explicit state duration. They capture non-local interactions through a joint sequence-structure probability distribution based on structural segments. In this paper, we investigate a combined architecture of Bayesian segmentation at the first stage and neural network at the second stage which captures both local and non-local correlation, to increase the single sequence prediction accuracy. The combined architecture is further enhanced by using neural network optimization and ensemble techniques.