Computer Sciences Commons^™

Classification For Mass Spectra And Comprehensive Two-Dimensional Chromatograms, Xue Tian

School of Computing: Dissertations, Theses, and Student Research

Mass spectra contain characteristic information regarding the molecular structure and properties of compounds. The mass spectra of compounds from the same chemically related group are similar. Classification is one of the fundamental methodologies for analyzing mass spectral data. The primary goals of classification are to automatically group compounds based on their mass spectra, to find correlation between the properties of compounds and their mass spectra, and to provide a positive identification of unknown compounds.

This dissertation presents a new algorithm for the classification of mass spectra, the most similar neighbor with a probability-based spectrum similarity measure (MSN-PSSM). Experimental results demonstrate …

Centinela: A Human Activity Recognition System Based On Acceleration And Vital Sign Data, Óscar D. Lara, Alfredo J. Perez, Miguel A. Labrador, José D. Posada

Computer Science Faculty Publications

This paper presents Centinela, a system that combines acceleration data with vital signs to achieve highly accurate activity recognition. Centinela recognizes five activities: walking, running, sitting, ascending, and descending. The system includes a portable and unobtrusive real-time data collection platform, which only requires a single sensing device and a mobile phone. To extract features, both statistical and structural detectors are applied, and two new features are proposed to discriminate among activities during periods of vital sign stabilization. After evaluating eight different classifiers and three different time window sizes, our results show that Centinela achieves up to 95.7% overall accuracy, which …

Double Updating Online Learning, Peilin Zhao, Steven C. H. Hoi, Rong Jin

Research Collection School Of Computing and Information Systems

In most kernel based online learning algorithms, when an incoming instance is misclassified, it will be added into the pool of support vectors and assigned with a weight, which often remains unchanged during the rest of the learning process. This is clearly insufficient since when a new support vector is added, we generally expect the weights of the other existing support vectors to be updated in order to reflect the influence of the added support vector. In this paper, we propose a new online learning method, termed Double Updating Online Learning, or DUOL for short, that explicitly addresses this problem. …

Processing And Classification Of Physiological Signals Using Wavelet Transform And Machine Learning Algorithms, Abed Al-Raoof Bsoul

Theses and Dissertations

Over the last century, physiological signals have been broadly analyzed and processed not only to assess the function of the human physiology, but also to better diagnose illnesses or injuries and provide treatment options for patients. In particular, Electrocardiogram (ECG), blood pressure (BP) and impedance are among the most important biomedical signals processed and analyzed. The majority of studies that utilize these signals attempt to diagnose important irregularities such as arrhythmia or blood loss by processing one of these signals. However, the relationship between them is not yet fully studied using computational methods. Therefore, a system that extract and combine …

Algorithms For Training Large-Scale Linear Programming Support Vector Regression And Classification, Pablo Rivas Perea

Open Access Theses & Dissertations

The main contribution of this dissertation is the development of a method to train a Support Vector Regression (SVR) model for the large-scale case where the number of training samples supersedes the computational resources. The proposed scheme consists of posing the SVR problem entirely as a Linear Programming (LP) problem and on the development of a sequential optimization method based on variables decomposition, constraints decomposition, and the use of primal-dual interior point methods. Experimental results demonstrate that the proposed approach has comparable performance with other SV-based classifiers. Particularly, experiments demonstrate that as the problem size increases, the sparser the solution …

Measuring Traffic Flow And Classifying Vehicle Types: A Surveillance Video Based Approach, Erhan İnce

Turkish Journal of Electrical Engineering and Computer Sciences

The paper presents a vehicle counting method based on invariant moments and shadow aware foreground masks. Estimation of the background and the segmentation of foreground regions can be done using either the Mixture of Gaussians model (MoG) or an improved version of the Group Based Histogram (GBH) technique. The work demonstrates that, even though the improved GBH method delivers performance just as good as MoG, considering computational efficiency, MoG is more suitable. Shadow aware binary masks for each frame are formed by using background subtraction and shadow removal in the Hue Saturation and Value (HSV) domain. To determine new vehicles …

Evolutionary Strategies For Data Mining, Rose Lowe

All Dissertations

Learning classifier systems (LCS) have been successful in generating rules for solving classification problems in data mining. The rules are of the form IF condition THEN action. The condition encodes the features of the input space and the action encodes the class label. What is lacking in those systems is the ability to express each feature using a function that is appropriate for that feature. The genetic algorithm is capable of doing this but cannot because only one type of membership function
is provided. Thus, the genetic algorithm learns only the shape and placement of the membership function, and in …

Comparative Analysis Of Combinations Of Dimension Reduction And Data Mining Techniques For Malware Detection, Proceso L. Fernandez Jr, Jeffrey C. Yiu, Paul Albert R. Arana

Department of Information Systems & Computer Science Faculty Publications

Many malware detectors utilize data mining techniques as primary tools for pattern recognition. As the number of new and evolving malware continues to rise, there is an increasing need for faster and more accurate detectors. However, for a given malware detector, detection speed and accuracy are usually inversely related. This study explores several configurations of classification combined with feature selection. An optimization function involving accuracy and processing time is used to evaluate each configuration. A real data set provided by Trend Micro Philippines is used for the study. Among 18 di↵erent configurations studied, it is shown that J4.8 without feature …

A Comparative Study Of Threshold-Based Feature Selection Techniques, Huanjing Wang, Taghi M. Khoshgoftaar, Jason Van Hulse

Computer Science Faculty Publications

Abstract Given high-dimensional software measurement data, researchers and practitioners often use feature (metric) selection techniques to improve the performance of software quality classification models. This paper presents our newly proposed threshold-based feature selection techniques, comparing the performance of these techniques by building classification models using five commonly used classifiers. In order to evaluate the effectiveness of different feature selection techniques, the models are evaluated using eight different performance metrics separately since a given performance metric usually captures only one aspect of the classification performance. All experiments are conducted on three Eclipse data sets with different levels of class imbalance. The …

The Impact Of Overfitting And Overgeneralization On The Classification Accuracy In Data Mining, Huy Nguyen Anh Pham

LSU Doctoral Dissertations

Current classification approaches usually do not try to achieve a balance between fitting and generalization when they infer models from training data. Such approaches ignore the possibility of different penalty costs for the false-positive, false-negative, and unclassifiable types. Thus, their performances may not be optimal or may even be coincidental. This dissertation analyzes the above issues in depth. It also proposes two new approaches called the Homogeneity-Based Algorithm (HBA) and the Convexity-Based Algorithm (CBA) to address these issues. These new approaches aim at optimally balancing the data fitting and generalization behaviors of models when some traditional classification approaches are used. …

Biomedical Relationship Extraction From Literature Based On Bio-Semantic Token Subsequences, Ying Xie, Jayasimha R. Katukuri, Vijay V. Raghavan

Faculty Articles

Relationship Extraction (RE) from biomedical literature is an important and challenging problem in both text mining and bioinformatics. Although various approaches have been proposed to extract protein?protein interaction types, their accuracy rates leave a large room for further exploring. In this paper, two supervised learning algorithms based on newly defined "bio-semantic token subsequence" are proposed for multi-class biomedical relationship classification. The first approach calculates a "bio-semantic token subsequence kernel", whereas the second one explicitly extracts weighted features from bio-semantic token subsequences. The two proposed approaches outperform several alternatives reported in literature on multi-class protein?protein interaction classification.

A Wrapper-Based Feature Selection For Analysis Of Large Data Sets, Jinsong Leng, Craig Valli, Leisa Armstrong

Research outputs pre 2011

Knowledge discovery from large data sets using classic data mining techniques has been proved to be difficult due to large size in both dimension and samples. In real applications, data sets often consist of many noisy, redundant, and irrelevant features, resulting in degrading the classification accuracy and increasing the complexity exponentially. Due to the inherent nature, the analysis of the quality of data sets is difficult and very limited approaches about this issue can be found in the literature. This paper presents a novel method to investigate the quality and structure of data sets, i.e., how to analyze whether there …

Towards Google Challenge: Combining Contextual And Social Information For Web Video Categorization, Xiao Wu, Wan-Lei Zhao, Chong-Wah Ngo

Research Collection School Of Computing and Information Systems

Web video categorization is a fundamental task for web video search. In this paper, we explore the Google challenge from a new perspective by combing contextual and social information under the scenario of social web. The semantic meaning of text (title and tags), video relevance from related videos, and user interest induced from user videos, are integrated to robustly determine the video category. Experiments on YouTube videos demonstrate the effectiveness of the proposed solution. The performance reaches 60% improvement compared to the traditional text based classifiers.

Temporal Data Classification Using Linear Classifiers, Peter Revesz, Thomas Triplet

School of Computing: Conference and Workshop Papers

Data classification is usually based on measurements recorded at the same time. This paper considers temporal data classification where the input is a temporal database that describes measurements over a period of time in history while the predicted class is expected to occur in the future. We describe a new temporal classification method that improves the accuracy of standard classification methods. The benefits of the method are tested on weather forecasting using the meteorological database from the Texas Commission on Environmental Quality.

A Novel Framework For Efficient Automated Singer Identification In Large Music Databases, Jialie Shen, John Shepherd, Bin Cui, Kian-Lee Tan

Research Collection School Of Computing and Information Systems

Over the past decade, there has been explosive growth in the availability of multimedia data, particularly image, video, and music. Because of this, content-based music retrieval has attracted attention from the multimedia database and information retrieval communities. Content-based music retrieval requires us to be able to automatically identify particular characteristics of music data. One such characteristic, useful in a range of applications, is the identification of the singer in a musical piece. Unfortunately, existing approaches to this problem suffer from either low accuracy or poor scalability. In this article, we propose a novel scheme, called Hybrid Singer Identifier (HSI), for …

Seasonal Adaptation Of Vegetation Color In Satellite Images For Flight Simulations, Yuzhong Shen, Jiang Li, Vamsi Mantena, Srinivas Jakkula

Electrical & Computer Engineering Faculty Publications

Automatic vegetation identification plays an important role in many applications including remote sensing and high performance flight simulations. This paper proposes a novel method that identifies vegetative areas in satellite images and then alters vegetation color to simulate seasonal changes based on training image pairs. The proposed method first generates a vegetation map for pixels corresponding to vegetative areas, using ISODATA clustering and vegetation classification. The ISODATA algorithm determines the number of clusters automatically. We then apply morphological operations to the clustered images to smooth the boundaries between clusters and to fill holes inside clusters. Six features are then computed …

A New Approach Using Temporal Radial Basis Function In Chronological Series, Mustapha Guezouri

Turkish Journal of Electrical Engineering and Computer Sciences

In this paper, we present an extended form of Radial Basis Function network called Temporal-RBF (T-RBF) network. This extended network can be used in decision rules and classification in Spatio-Temporal domain applications, like speech recognition, economic fluctuations, seismic measurements and robotics applications. We found that such a network complies with relative ease to constraints such as capacity of universal approximation, sensibility of node, local generalisation in receptive field, etc. For an optimal solution based on a probabilistic approach with a minimum of complexity, we propose two TRBF models (1 and 2). Application to the problem of Mackey-Glass time series has …

An Automatic Bridge Detection Technique For Multispectral Images, D. Chaudhuri, Ashok Samal

School of Computing: Conference and Workshop Papers

Extraction of features from images has been a goal of researchers since the early days of remote sensing. While significant progress has been made in several applications, much remains to be done in the area of accurate identification of high-level features such as buildings and roads. This paper presents an approach for detecting bridges over water bodies from multispectral imagery. The multispectral image is first classified into eight land-cover types using a majority-must-be-granted logic based on the multiseed supervised classification technique. The classified image is then categorized into a trilevel image: water, concrete, and background. Bridges are then recognized in …

Sentiment Regression: Using Real-Valued Scores To Summarize Overall Document Sentiment, Adam Drake, Eric K. Ringger, Dan A. Ventura

Faculty Publications

In this paper, we consider a sentiment regression problem: summarizing the overall sentiment of a review with a real-valued score. Empirical results on a set of labeled reviews show that real-valued sentiment modeling is feasible, as several algorithms improve upon baseline performance. We also analyze performance as the granularity of the classification problem moves from two-class (positive vs. negative) towards infinite-class (real-valued).

Vegetation Identification Based On Satellite Imagery, Vamsi K.R. Mantena, Ramu Pedada, Srinivas Jakkula, Yuzhong Shen, Jiang Li, Hamid R. Arabnia (Ed.)

Electrical & Computer Engineering Faculty Publications

Automatic vegetation identification plays an important role in many applications including remote sensing and high performance flight simulations. This paper presents a method to automatically identify vegetation based upon satellite imagery. First, we utilize the ISODATA algorithm to cluster pixels in the images where the number of clusters is determined by the algorithm. We then apply morphological operations to the clustered images to smooth the boundaries between clusters and to fill holes inside clusters. After that, we compute six features for each cluster. These six features then go through a feature selection algorithm and three of them are determined to …

Learning Multiple Languages In Groups, Sanjay Jain, Efim Kinber

School of Computer Science & Engineering Faculty Publications

We consider a variant of Gold’s learning paradigm where a learner receives as input different languages (in the form of one text where all input languages are interleaved). Our goal is to explore the situation when a more “coarse” classification of input languages is possible, whereas more refined classification is not. More specifically, we answer the following question: under which conditions, a learner, being fed different languages, can produce grammars covering all input languages, but cannot produce grammars covering input languages for any . We also consider a variant of this task, where each of the output grammars may not …

Improving Neural Network Classification Training, Michael Edwin Rimer

Theses and Dissertations

The following work presents a new set of general methods for improving neural network accuracy on classification tasks, grouped under the label of classification-based methods. The central theme of these approaches is to provide problem representations and error functions that more directly improve classification accuracy than conventional learning and error functions. The CB1 algorithm attempts to maximize classification accuracy by selectively backpropagating error only on misclassified training patterns. CB2 incorporates a sliding error threshold to the CB1 algorithm, interpolating between the behavior of CB1 and standard error backpropagation as training progresses in order to avoid prematurely saturated network weights. CB3 …

Data Mining And Analysis On Multiple Time Series Object Data, Chunyu Jiang

Browse all Theses and Dissertations

Huge amount of data is available in our society and the need for turning such data into useful information and knowledge is urgent. Data mining is an important field addressing that need and significant progress has been achieved in the last decade. In several important application areas, data arises in the format of Multiple Time Series Object (MTSO) data, where each data object is an array of time series over a large set of features and each has an associated class or state. Very little research has been conducted towards this kind of data. Examples include computational toxicology, where each …

Exploration Of Computational Methods For Classification Of Movement Intention During Human Voluntary Movement From Single Trial Eeg, Ou Bai, Peter Lin, Sherry Vorbach, Jiang Li, Steve Furlani, Mark Hallett

Electrical & Computer Engineering Faculty Publications

Objective: To explore effective combinations of computational methods for the prediction of movement intention preceding the production of self-paced right and left hand movements from single trial scalp electroencephalogram (EEG).

Methods: Twelve naïve subjects performed self-paced movements consisting of three key strokes with either hand. EEG was recorded from 128 channels. The exploration was performed offline on single trial EEG data. We proposed that a successful computational procedure for classification would consist of spatial filtering, temporal filtering, feature selection, and pattern classification. A systematic investigation was performed with combinations of spatial filtering using principal component analysis (PCA), independent component analysis …

Towards Effective Content-Based Music Retrieval With Multiple Acoustic Feature Combination, Jialie Shen, John Shepherd, Ann H. H. Ngu

Research Collection School Of Computing and Information Systems

In this paper, we present a new approach to constructing music descriptors to support efficient content-based music retrieval and classification. The system applies multiple musical properties combined with a hybrid architecture based on principal component analysis (PCA) and a multilayer perceptron neural network. This architecture enables straightforward incorporation of multiple musical feature vectors, based on properties such as timbral texture, pitch, and rhythm structure, into a single low-dimensioned vector that is more effective for classification than the larger individual feature vectors. The use of supervised training enables incorporation of human musical perception that further enhances the classification process. We compare …

A Model For Anticipatory Event Detection, Qi He, Kuiyu Chang, Ee Peng Lim

Research Collection School Of Computing and Information Systems

Event detection is a very important area of research that discovers new events reported in a stream of text documents. Previous research in event detection has largely focused on finding the first story and tracking the events of a specific topic. A topic is simply a set of related events defined by user supplied keywords with no associated semantics and little domain knowledge. We therefore introduce the Anticipatory Event Detection (AED) problem: given some user preferred event transition in a topic, detect the occurence of the transition for the stream of news covering the topic. We confine the events to …

Learning The Unified Kernel Machines For Classification, Steven C. H. Hoi, Michael R. Lyu, Edward Y. Chang

Research Collection School Of Computing and Information Systems

Kernel machines have been shown as the state-of-the-art learning techniques for classification. In this paper, we propose a novel general framework of learning the Unified Kernel Machines (UKM) from both labeled and unlabeled data. Our proposed framework integrates supervised learning, semi-supervised kernel learning, and active learning in a unified solution. In the suggested framework, we particularly focus our attention on designing a new semi-supervised kernel learning method, i.e., Spectral Kernel Learning (SKL), which is built on the principles of kernel target alignment and unsupervised kernel design. Our algorithm is related to an equivalent quadratic programming problem that can be efficiently …

Fisa: Feature-Based Instance Selection For Imbalanced Text Classification, Aixin Sun, Ee Peng Lim, Boualem Benatallah, Mahbub Hassan

Research Collection School Of Computing and Information Systems

Support Vector Machines (SVM) classifiers are widely used in text classification tasks and these tasks often involve imbalanced training. In this paper, we specifically address the cases where negative training documents significantly outnumber the positive ones. A generic algorithm known as FISA (Feature-based Instance Selection Algorithm), is proposed to select only a subset of negative training documents for training a SVM classifier. With a smaller carefully selected training set, a SVM classifier can be more efficiently trained while delivering comparable or better classification accuracy. In our experiments on the 20-Newsgroups dataset, using only 35% negative training examples and 60% learning …

Webarc: Website Archival Using A Structured Approach, Ee Peng Lim, Maria Marissa

Research Collection School Of Computing and Information Systems

Website archival refers to the task of monitoring and storing snapshots of website(s) for future retrieval and analysis. This task is particularly important for websites that have content changing over time with older information constantly overwritten by newer one. In this paper, we propose WEBARC as a set of software tools to allow users to construct a logical structure for a website to be archived. Classifiers are trained to. determine relevant web pages and their categories, and subsequently used in website downloading. The archival schedule can be specified and executed by a scheduler. A website viewer is also developed to …

Translation Initiation Sites Prediction With Mixture Gaussian Models In Human Cdna Sequences, G. Li, Tze-Yun Leong, Louxin Zhang

Research Collection School Of Computing and Information Systems

Translation initiation sites (TISs) are important signals in cDNA sequences. Many research efforts have tried to predict TISs in cDNA sequences. In this paper, we propose to use mixture Gaussian models for TIS prediction. Using both local features and some features generated from global measures, the proposed method predicts TISs with a sensitivity of 98 percent and a specificity of 93.6 percent. Our method outperforms many other existing methods in sensitivity while keeping specificity high. We attribute the improvement in sensitivity to the nature of the global features and the mixture Gaussian models. © 2005 IEEE.