Physical Sciences and Mathematics Commons^™

Mining Weakly Labeled Web Facial Images For Search-Based Face Annotation, Dayong Wang, Steven C. H. Hoi, Ying He, Jianke Zhu

Research Collection School Of Computing and Information Systems

This paper investigates a framework of search-based face annotation (SBFA) by mining weakly labeled facial images that are freely available on the World Wide Web (WWW). One challenging problem for search-based face annotation scheme is how to effectively perform annotation by exploiting the list of most similar facial images and their weak labels that are often noisy and incomplete. To tackle this problem, we propose an effective unsupervised label refinement (ULR) approach for refining the labels of web facial images using machine learning techniques. We formulate the learning problem as a convex optimization and develop effective optimization algorithms to solve …

Online Portfolio Selection: A Survey, Bin Li, Steven C. H. Hoi

Research Collection School Of Computing and Information Systems

Online portfolio selection is a fundamental problem in computational finance, which has been extensively studied across several research communities, including finance, statistics, artificial intelligence, machine learning, and data mining. This article aims to provide a comprehensive survey and a structural understanding of online portfolio selection techniques published in the literature. From an online machine learning perspective, we first formulate online portfolio selection as a sequential decision problem, and then we survey a variety of state-of-the-art approaches, which are grouped into several major categories, including benchmarks, Follow-the-Winner approaches, Follow-the-Loser approaches, Pattern-Matching--based approaches, and Meta-Learning Algorithms. In addition to the problem formulation …

Can Clustering Improve Requirements Traceability? A Tracelab-Enabled Study, Brett Taylor Armstrong

Master's Theses

Software permeates every aspect of our modern lives. In many applications, such in the software for airplane flight controls, or nuclear power control systems software failures can have catastrophic consequences. As we place so much trust in software, how can we know if it is trustworthy? Through software assurance, we can attempt to quantify just that.

Building complex, high assurance software is no simple task. The difficult information landscape of a software engineering project can make verification and validation, the process by which the assurance of a software is assessed, very difficult. In order to manage the inevitable information overload …

Using Machine Learning Techniques To Customize The User's Profile, Helps Intelligent Tv Decoder’S Design, Alketa Hyso, Roneda Mucaj

UBT International Conference

In today's society due to the increase of the quantity of information is becoming more difficult to find the information we search. "Data mining" offers us the most important methods and techniques in data analysis. Through this work, we aim to study the several data mining techniques, methods and applications in specific areas. We experiment with an “open software" WEKA, to perform some data analysis, presenting the reliability and advantages of data mining classification technique. We use the decision trees technique to achieve the task of classification, to customize user profiles based on their requirements and needs. This paper presents …

What You Want Is Not What You Get: Predicting Sharing Policies For Text-Based Content On Facebook, Arunesh Sinha, Li Yan, Lujo Bauer

Research Collection Lee Kong Chian School Of Business

As the amount of content users publish on social networking sites rises, so do the danger and costs of inadvertently sharing content with an unintended audience. Studies repeatedly show that users frequently misconfigure their policies or misunderstand the privacy features offered by social networks. A way to mitigate these problems is to develop automated tools to assist users in correctly setting their policy. This paper explores the viability of one such approach: we examine the extent to which machine learning can be used to deduce users' sharing preferences for content posted on Facebook. To generate data on which to evaluate …

Variable Importance And Prediction Methods For Longitudinal Problems With Missing Variables, Ivan Diaz, Alan E. Hubbard, Anna Decker, Mitchell Cohen

U.C. Berkeley Division of Biostatistics Working Paper Series

In this paper we present prediction and variable importance (VIM) methods for longitudinal data sets containing both continuous and binary exposures subject to missingness. We demonstrate the use of these methods for prognosis of medical outcomes of severe trauma patients, a field in which current medical practice involves rules of thumb and scoring methods that only use a few variables and ignore the dynamic and high-dimensional nature of trauma recovery. Well-principled prediction and VIM methods can thus provide a tool to make care decisions informed by the high-dimensional patient’s physiological and clinical history. Our VIM parameters can be causally interpreted …

Enabling Richer Insight Into Runtime Executions Of Systems, Karthik Swaminathan Nagaraj

Open Access Dissertations

Systems software of very large scales are being heavily used today in various important scenarios such as online retail, banking, content services, web search and social networks. As the scale of functionality and complexity grows in these software, managing the implementations becomes a considerable challenge for developers, designers and maintainers. Software needs to be constantly monitored and tuned for optimal efficiency and user satisfaction. With large scale, these systems incorporate significant degrees of asynchrony, parallelism and distributed executions, reducing the manageability of software including performance management. Adding to the complexity, developers are under pressure between developing new functionality for customers …

Asymptotically Unbiased Estimator Of The Informational Energy With Knn, Angel Caţaron, Răzvan Andonie, Chinmei Y. Chueh

All Faculty Scholarship for the College of the Sciences

Motivated by machine learning applications (e.g., classification, function approximation, feature extraction), in previous work, we have introduced a non- parametric estimator of Onicescu’s informational energy. Our method was based on the k-th nearest neighbor distances between the n sample points, where k is a fixed positive integer. In the present contribution, we discuss mathematical properties of this estimator. We show that our estimator is asymptotically unbiased and consistent. We provide further experimental results which illustrate the convergence of the estimator for standard distributions.

Practical Cost-Conscious Active Learning For Data Annotation In Annotator-Initiated Environments, Robbie A. Haertel

Theses and Dissertations

Many projects exist whose purpose is to augment raw data with annotations that increase the usefulness of the data. The number of these projects is rapidly growing and in the age of “big data” the amount of data to be annotated is likewise growing within each project. One common use of such data is in supervised machine learning, which requires labeled data to train a predictive model. Annotation is often a very expensive proposition, particularly for structured data. The purpose of this dissertation is to explore methods of reducing the cost of creating such data sets, including annotated text corpora.We …

Segmentation And Model Generation For Large-Scale Cyber Attacks, Steven E. Strapp

Theses

Raw Cyber attack traffic can present more questions than answers to security analysts. Especially with large-scale observables it is difficult to identify which packets are relevant and what attack behaviors are present. Many existing works in Host or Flow Clustering attempt to group similar behaviors to expedite analysis; these works often phrase the problem directly as offline unsupervised machine learning. This work proposes online processing to simultaneously model coordinating actors and segment traffic that is relevant to a target of interest, all while it is being received. The goal is not just to aggregate similar attack behaviors, but to provide …

Computer Sketch Recognition, Richard Steigerwald

Master's Theses

Tens of thousands of years ago, humans drew sketches that we can see and identify even today. Sketches are the oldest recorded form of human communication and are still widely used. The universality of sketches supersedes that of culture and language. Despite the universal accessibility of sketches by humans, computers are unable to interpret or even correctly identify the contents of sketches drawn by humans with a practical level of accuracy.

In my thesis, I demonstrate that the accuracy of existing sketch recognition techniques can be improved by optimizing the classification criteria. Current techniques classify a 20,000 sketch crowd-sourced dataset …

Mind Change Speed-Up For Learning Languages From Positive Data, Sanjay Jain, Efim Kinber

School of Computer Science & Engineering Faculty Publications

Within the frameworks of learning in the limit of indexed classes of recursive languages from positive data and automatic learning in the limit of indexed classes of regular languages (with automatically computable sets of indices), we study the problem of minimizing the maximum number of mind changes by a learner on all languages with indices not exceeding . For inductive inference of recursive languages, we establish two conditions under which can be made smaller than any recursive unbounded non-decreasing function. We also establish how is affected if at least one of these two conditions does not hold. In the case …

Document Classification, Shane K. Panter

Boise State University Theses and Dissertations

We present an overview of the document classification process and present research conducted against the newly constructed SBIR-STTR corpus. Specifically, the current methods in use for annotation, corpus construction, feature construction, feature weighting, and classifier algorithms are surveyed. We introduce a new dataset derived from public data downloaded from sbir.gov and the Text Annotation Toolkit (TAT) ₁ for use in classification research.

TAT is a collection of independent components packaged together into one open source software application. TAT was engineered to support the document classification process and workflow. Tracking of changes in a working corpus, saving data used in the …

Probabilistic Explicit Topic Modeling, Joshua Aaron Hansen

Theses and Dissertations

Latent Dirichlet Allocation (LDA) is widely used for automatic discovery of latent topics in document corpora. However, output from analysis using an LDA topic model suffers from a lack of identifiability between topics not only across corpora, but across runs of the algorithm. The output is also isolated from enriching information from knowledge sources such as Wikipedia and is difficult for humans to interpret due to a lack of meaningful topic labels. This thesis introduces two methods for probabilistic explicit topic modeling that address these issues: Latent Dirichlet Allocation with Static Topic-Word Distributions (LDA-STWD), and Explicit Dirichlet Allocation (EDA). LDA-STWD …

Artificial Immune Systems And Particle Swarm Optimization For Solutions To The General Adversarial Agents Problem, Jeremy Mange

Dissertations

The general adversarial agents problem is an abstract problem description touching on the fields of Artificial Intelligence, machine learning, decision theory, and game theory. The goal of the problem is, given one or more mobile agents, each identified as either “friendly" or “enemy", along with a specified environment state, to choose an action or series of actions from all possible valid choices for the next “timestep" or series thereof, in order to lead toward a specified outcome or set of outcomes. This dissertation explores approaches to this problem utilizing Artificial Immune Systems, Particle Swarm Optimization, and hybrid approaches, along with …

Object Detection And Recognition In Natural Settings, George William Dittmar

Dissertations and Theses

Much research as of late has focused on biologically inspired vision models that are based on our understanding of how the visual cortex processes information. One prominent example of such a system is HMAX [17]. HMAX attempts to simulate the biological process for object recognition in cortex based on the model proposed by Hubel & Wiesel [10]. This thesis investigates the ability of an HMAX-like system (GLIMPSE [20]) to perform object-detection in cluttered natural scenes. I evaluate these results using the StreetScenes database from MIT [1, 8]. This thesis addresses three questions: (1) Can the GLIMPSE-based object detection system replicate …

An Automated Prognosis System For Estrogen Hormone Status Assessment In Breast Cancer Tissue Samples, Fati̇h Sarikoç, Adem Kalinli, Hülya Akgün, Fi̇gen Öztürk

Turkish Journal of Electrical Engineering and Computer Sciences

Estrogen receptor (ER) status evaluation is a widely applied method in the prognosis of breast cancer. However, testing for the existence of the ER biomarker in a patient's tumor sample mainly depends on the subjective decisions of the doctors. The aim of this paper is to introduce the usage of a machine learning tool, functional trees (FTs), to attain an ER prognosis of the disease via an objective decision model. For this aim, 27 image files, each of which came from a biopsy sample of an invasive ductal carcinoma patient, were scanned and captured by a light microscope. From these …

Artificial Intelligence And Data Mining: Algorithms And Applications, Jianhong Xia, Fuding Xie, Yong Zhang, Craig Caulfield

Research outputs 2013

Artificial intelligence and data mining techniques have been used in many domains to solve classification, segmentation, association, diagnosis, and prediction problems. The overall aim of this special issue is to open a discussion among researchers actively working on algorithms and applications. The issue covers a wide variety of problems for computational intelligence, machine learning, time series analysis, remote sensing image mining, and pattern recognition. After a rigorous peer review process, 20 papers have been selected from 38 submissions. The accepted papers in this issue addressed the following topics: (i) advanced artificial intelligence and data mining techniques; (ii) computational intelligence in …

A Machine Learning Approach To Diagnosis Of Parkinson’S Disease, Sumaiya F. Hashmi

CMC Senior Theses

I will investigate applications of machine learning algorithms to medical data, adaptations of differences in data collection, and the use of ensemble techniques.

Focusing on the binary classification problem of Parkinson’s Disease (PD) diagnosis, I will apply machine learning algorithms to a primary dataset consisting of voice recordings from healthy and PD subjects. Specifically, I will use Artificial Neural Networks, Support Vector Machines, and an Ensemble Learning algorithm to reproduce results from [MS12] and [GM09].

Next, I will adapt a secondary regression dataset of PD recordings and combine it with the primary binary classification dataset, testing various techniques to consolidate …

Human Intention Recognition Based Assisted Telerobotic Grasping Of Objects In An Unstructured Environment, Karan Hariharan Khokar

USF Tampa Graduate Theses and Dissertations

In this dissertation work, a methodology is proposed to enable a robot to identify an object to be grasped and its intended grasp configuration while a human is teleoperating a robot towards the desired object. Based on the detected object and grasp configuration, the human is assisted in the teleoperation task. The environment is unstructured and consists of a number of objects, each with various possible grasp configurations. The identification of the object and the grasp configuration is carried out in real time, by recognizing the intention of the human motion. Simultaneously, the human user is assisted to preshape over …

Automated Detection Of Vehicles With Machine Learning, Michael N. Johnstone, Andrew Woodward

Australian Information Security Management Conference

Considering the significant volume of data generated by sensor systems and network hardware which is required to be analysed and intepreted by security analysts, the potential for human error is significant. This error can lead to consequent harm for some systems in the event of an adverse event not being detected. In this paper we compare two machine learning algorithms that can assist in supporting the security function effectively and present results that can be used to select the best algorithm for a specific domain. It is suggested that a naive Bayesian classiifer (NBC) and an artificial neural network (ANN) …

Anticipating The Friction Coefficient Of Friction Materials Used In Automobiles By Means Of Machine Learning Without Using A Test Instrument, Mustafa Ti̇mur, Fati̇h Aydin

Turkish Journal of Electrical Engineering and Computer Sciences

The most important factor for designs in which friction materials are used is the coefficient of friction. The coefficient of friction has been determined taking such variants as velocity, temperature, and pressure into account, which arise from various factors in friction materials, and by analyzing the effects of these variants on friction materials. Many test instruments have been produced in order to determine the coefficient of friction. In this article, a study about the use of machine learning algorithms instead of test instruments in order to determine the coefficient of friction is presented. Isotonic regression was selected as the machine …

Concept Drift Datasets, Patrick Lindstrom

Doctoral

This zip file contains the datasets used in the PhD thesis:

Lindstrom, P., 2013. Handling Concept Drift in the Context of Expensive Labels. Technological University Dublin. For more information about the datasets please see the README file and the aforementioned thesis.

On Identifying Critical Nuggets Of Information During Classification Task, David Sathiaraj

LSU Doctoral Dissertations

In large databases, there may exist critical nuggets - small collections of records or instances that contain domain-specific important information. This information can be used for future decision making such as labeling of critical, unlabeled data records and improving classification results by reducing false positive and false negative errors. In recent years, data mining efforts have focussed on pattern and outlier detection methods. However, not much effort has been dedicated to finding critical nuggets within a data set. This work introduces the idea of critical nuggets, proposes an innovative domain-independent method to measure criticality, suggests a heuristic to reduce the …

Computationally Efficient Confidence Intervals For Cross-Validated Area Under The Roc Curve Estimates, Erin Ledell, Maya L. Petersen, Mark J. Van Der Laan

U.C. Berkeley Division of Biostatistics Working Paper Series

In binary classification problems, the area under the ROC curve (AUC), is an effective means of measuring the performance of your model. Most often, cross-validation is also used, in order to assess how the results will generalize to an independent data set. In order to evaluate the quality of an estimate for cross-validated AUC, we must obtain an estimate for its variance. For massive data sets, the process of generating a single performance estimate can be computationally expensive. Additionally, when using a complex prediction method, calculating the cross-validated AUC on even a relatively small data set can still require a …

Identification Of Tcp Protocols, Juan Shao

School of Computing: Dissertations, Theses, and Student Research

Recently, many new TCP algorithms, such as BIC, CUBIC, and CTCP, have been deployed in the Internet. Investigating the deployment statistics of these TCP algorithms is meaningful to study the performance and stability of the Internet. Currently, there is a tool named Congestion Avoidance Algorithm Identification (CAAI) for identifying the TCP algorithm of a web server and then for investigating the TCP deployment statistics. However, CAAI using a simple k-NN algorithm can not achieve a high identification accuracy. In this thesis, we comprehensively study the identification accuracy of five popular machine learning models. We find that the random forest model …

Bayesian Test Analytics For Document Collections, Daniel David Walker

Theses and Dissertations

Modern document collections are too large to annotate and curate manually. As increasingly large amounts of data become available, historians, librarians and other scholars increasingly need to rely on automated systems to efficiently and accurately analyze the contents of their collections and to find new and interesting patterns therein. Modern techniques in Bayesian text analytics are becoming wide spread and have the potential to revolutionize the way that research is conducted. Much work has been done in the document modeling community towards this end,though most of it is focused on modern, relatively clean text data. We present research for improved …

A Confidence-Prioritization Approach To Data Processing In Noisy Data Sets And Resulting Estimation Models For Predicting Streamflow Diel Signals In The Pacific Northwest, Nathaniel Lee Gustafson

Theses and Dissertations

Streams in small watersheds are often known to exhibit diel fluctuations, in which streamflow oscillates on a 24-hour cycle. Streamflow diel fluctuations, which we investigate in this study, are an informative indicator of environmental processes. However, in Environmental Data sets, as well as many others, there is a range of noise associated with individual data points. Some points are extracted under relatively clear and defined conditions, while others may include a range of known or unknown confounding factors, which may decrease those points' validity. These points may or may not remain useful for training, depending on how much uncertainty they …

On The K-Mer Frequency Spectra Of Organism Genome And Proteome Sequences With A Preliminary Machine Learning Assessment Of Prime Predictability, Nathan O. Schmidt

Boise State University Theses and Dissertations

A regular expression and region-specific filtering system for biological records at the National Center for Biotechnology database is integrated into an object oriented sequence counting application, and a statistical software suite is designed and deployed to interpret the resulting k-mer frequencies|with a priority focus on nullomers. The proteome k-mer frequency spectra of ten model organisms and the genome k-mer frequency spectra of two bacteria and virus strains for the coding and non-coding regions are comparatively scrutinized. We observe that the naturally-evolved (NCBI/organism) and the artificially-biased (randomly-generated) sequences exhibit a clear deviation from the artificially-unbiased (randomly-generated) histogram distributions. …

On The Automatic Recognition Of Human Activities Using Heterogeneous Wearable Sensors, Oscar David Lara Yejas

USF Tampa Graduate Theses and Dissertations

Delivering accurate and opportune information on people's activities and behaviors has become one of the most important tasks within pervasive computing. Its wide spectrum of potential applications in medical, entertainment, and tactical scenarios, motivates further

research and development of new strategies to improve accuracy, pervasiveness, and eciency.

This dissertation addresses the recognition of human activities (HAR) with wearable sensors in three main regards: In the rst place, physiological signals have been incorporated as a new source of information to improve the recognition accuracy achieved by conventional approaches, which rely on accelerometer signals solely. A new HAR system, Centinela, was born …