Search

We propose a structural health monitoring framework that is based on the hybrid intelligent systems to compute composite structure health index as structure health indicator. With our proposed health structure monitor framework, it provides a decision-level analysis for structure health monitoring. In contrast, recent studies focus on signal processing, mechanical modelling and computer based systems, which are not robust enough to provide decision-level analysis confronted with heterogeneous data sources. Therefore, data fusion technique is required to alleviate possibilities of data conflicts, lack of conciseness and incompleteness that arise from the heterogeneous data sources to compute composite structure health index, which is essential for risk analysis. Our proposed intelligent data fusion framework is based on hybrid adaptive resonant theory neural network (superior learning from dynamic data) and adaptive fuzzy inference (superior reasoning with dynamic fuzzy rules derived from time series data formulated membership functions) systems. The outcome of our proposed framework is a structure composite health index, an aggregated index via optimal weighted variables. We implement our framework using a set of GPS and Wind velocity sensors output from a real bridge to evaluate our framework. These sensors data used in our study are collected from long cable-bridge. Our case study showed that our proposed intelligent data fusing framework is capable of reporting the structural status correctly.

This paper proposes an innovative fraud detection method, built upon existing fraud detection research and Minority Report , to deal with the data mining problem of skewed data distributions. This method uses backpropagation (BP), together with naive Bayesian (NB) and C4.5 algorithms, on data partitions derived from minority oversampling with replacement. Its originality lies in the use of a single meta-classifier (stacking) to choose the best base classifiers, and then combine these base classifiers' predictions (bagging) to improve cost savings (stacking-bagging). Results from a publicly available automobile insurance fraud detection data set demonstrate that stacking-bagging performs slightly better than the best performing bagged algorithm, C4.5, and its best classifier, C4.5 (2), in terms of cost savings. Stacking-bagging also outperforms the common technique used in industry (BP without both sampling and partitioning). Subsequently, this paper compares the new fraud detection method (meta-learning approach) against C4.5 trained using undersampling, oversampling, and SMOTEing without partitioning (sampling approach). Results show that, given a fixed decision threshold and cost matrix, the partitioning and multiple algorithms approach achieves marginally higher cost savings than varying the entire training data set with different class distributions. The most interesting find is confirming that the combination of classifiers to produce the best cost savings has its contributions from all three algorithms.

This paper proposes and tests a methodology for selecting features and test cases with the goal of improving medium term bankruptcy prediction accuracy in large uncontrolled datasets of financial records. We propose a Genetic Programming and Neural Network based objective feature selection methodology to identify key inputs, and then use those inputs to combine multi-level Self-Organising Maps with Spectral Clustering to build clusters. Performing objective feature selection within each of those clusters, this research was able to increase out-of-sample classification accuracy from 71.3% and 69.8% on the Genetic Programming and Neural Network models respectively to 80.0% and 77.3%.

We present a system to detect parked vehicles in a typical commercial parking complex using multiple streams of images captured through IP connected devices. Compared to traditional object detection techniques and machine learning methods, our approach is significantly faster in detection speed in the presence of multiple image streams. It is also capable of comparable accuracy when put to test against existing methods. And this is achieved without the need to train the system that machine learning methods require. Our approach uses a combination of psychological insights obtained from human detection and an algorithm replicating the outcomes of a SVM learner but without the noise that compromises accuracy in the normal learning process. The result is faster detection with comparable accuracy. Our experiments on images captured from a local test site shows very promising results for an implementation that is not only effective and low cost but also opens doors to new parking applications when combined with other technologies.

Timely detection of hidden patterns is the key for the analysis and estimating of driving determinants for mission critical decision making. This study applies Cheong and Lee’s “context-aware” content analysis framework to extract latent properties from Twitter messages (tweets). In addition, we incorporate an unsupervised Self-organizing Feature Map (SOM) as a machine learning-based clustering tool that has not been investigated in the context of opinion mining and sentimental analysis using microblogging. Our experimental results reveal the detection of interesting patterns for topics of interest which are latent and cannot be easily detected from the observed tweets without the aid of machine learning tools.

This paper provides an explanation to how a recently developed machine learning approach, namely the Probability Measure Graph Self-Organizing Map (PM-GraphSOM) can be used for the generation of links between referenced or otherwise interlinked documents. This new generation of SOM models are capable of projecting generic graph structured data onto a fixed sized display space. Such a mechanism is normally used for dimension reduction, visualization, or clustering purposes. This paper shows that the PM-GraphSOM training algorithm "inadvertently" encodes relations that exist between the atomic elements in a graph. If the nodes in the graph represent documents, and the links in the graph represent the reference (or hyperlink) structure of the documents, then it is possible to obtain a set of links for a test document whose link structure is unknown. A significant finding of this paper is that the described approach is scalable in that links can be extracted in linear time. It will also be shown that the proposed approach is capable of predicting the pages which would be linked to a new document, and is capable of predicting the links to other documents from a given test document. The approach is applied to web pages from Wikipedia, a relatively large XML text database consisting of many referenced documents.

Cluster analysis has played a key role in data understanding. When such an important data mining task is extended to the context of data streams, it becomes more challenging since the data arrive at a mining system in one-pass manner. The problem is even more difficult when the clustering task is considered in a sliding window model which requiring the elimination of outdated data must be dealt with properly. We propose SWEM algorithm that exploits the Expectation Maximization technique to address these challenges. SWEM is not only able to process the stream in an incremental manner, but also capable to adapt to changes happened in the underlying stream distribution.

In this chapter, we will describe a method for extracting an underlying graph structure from an unstructured text document. The resulting graph structure is a symmetrical un-directed graph. An unsupervised learning approach is applied to cluster a given text corpus into groups of similar structured graphs. Moreover, if labels are given to some of the documents in the text corpus, a supervised learning approach can be applied to learn the underlying input-output mapping between the symmetrical un-directed graph structures and a real-valued vector. The approach will be illustrated using a standard benchmark problem in text processing, viz., a subset of the Reuters text corpus. Some observations and further research directions are given.

Gradient descent is a widely used paradigm for solving many optimization problems. Stochastic gradient descent performs a series of iterations to minimize a target function in order to reach a local minimum. In machine learning or data mining, this function corresponds to a decision model that is to be discovered. The gradient descent paradigm underlies many commonly used techniques in data mining and machine learning, such as neural networks, Bayesian networks, genetic algorithms, and simulated annealing. To the best of our knowledge, there has not been any work that extends the notion of privacy preservation or secure multi-party computation to gradient-descent-based techniques. In this paper, we propose a preliminary approach to enable privacy preservation in gradient descent methods in general and demonstrate its feasibility in specific gradient descent methods.

In retail industry, it is very important to understand seasonal sales pattern, because this knowledge can assist decision makers in managing inventory and formulating marketing strategies. Self-Organizing Map (SOM) is suitable for extracting and illustrating essential structures because SOM has unsupervised learning and topology preserving properties, and prominent visualization techniques. In this experiment, we propose a method for seasonal pattern analysis using Self-Organizing Map. Performance test with real-world data from stationery stores in Indonesia shows that the method is effective for seasonal pattern analysis. The results are used to formulate several marketing and inventory management strategies.

We exploit the merits of C4.5 decision tree classifier with two stacking meta-learners: back-propagation multilayer perceptron neural network and naive-Bayes respectively The performance of these two hybrid classification schemes have been empirically tested and compared with C4.5 decision tree using two US data sets (raw data set and new data set incorporated with domain knowledge) simultaneously to predict US bank failure Significant improvements in prediction accuracy and training efficiency have been achieved in the schemes based on new data set The empirical test results suggest that the proposed hybrid schemes perform marginally better in term of AUC criterion.

Knowledge discovery from massive business data is a nontrivial research issue. A generalized framework to guide knowledge discovery process is necessary to improve its efficiency and effectiveness. This paper proposes a framework to relate certainty factor to an absolute factor hereafter called alpha (α) factor. Alpha factor represents the magnitude of useful local knowledge that is extracted from raw data in a fuzzy inference system also its correctness in relate to a global knowledge. The concept of alpha, α, is explained with mathematical illustration and a research design. Three specific case studies are included to illustrate the use of the proposed self-learning framework for true discovery.

The objective is to measure utility of real-time commercial decision making. It is important due to a higher possibility of mistakes in real-time decisions, problems with recording actual occurrences, and significant costs associated with predictions produced by algorithms. The first contribution is to use overall utility and represent individual utility with a monetary value instead of a prediction. The second is to calculate the benefit from predictions using the utility-based decision threshold. The third is to incorporate cost of predictions. For experiments, overall utility is used to evaluate communal and spike detection, and their adaptive versions. The overall utility results show that with fewer alerts, communal detection is better than spike detection. With more alerts, adaptive communal and spike detection are better than their static versions. To maximise overall utility with all algorithms, only 1% to 4% in the highest predictions should be alerts.