nach oben

2010 | Buch

Kapitel lesen Erstes Kapitel lesen

Advances in Machine Learning and Data Analysis

herausgegeben von: Mahyar A. Amouzegar

Verlag: Springer Netherlands

Buchreihe : Lecture Notes in Electrical Engineering

Enthalten in: Springer Professional "Wirtschaft+Technik" , Springer Professional "Technik" , Springer Professional "Wirtschaft"

Einloggen, um Zugang zu erhalten

Über dieses Buch

A large international conference on Advances in Machine Learning and Data Analysis was held in UC Berkeley, California, USA, October 22-24, 2008, under the auspices of the World Congress on Engineering and Computer Science (WCECS 2008). This volume contains sixteen revised and extended research articles written by prominent researchers participating in the conference. Topics covered include Expert system, Intelligent decision making, Knowledge-based systems, Knowledge extraction, Data analysis tools, Computational biology, Optimization algorithms, Experiment designs, Complex system identification, Computational modeling, and industrial applications. Advances in Machine Learning and Data Analysis offers the state of the art of tremendous advances in machine learning and data analysis and also serves as an excellent reference text for researchers and graduate students, working on machine learning and data analysis.

Inhaltsverzeichnis

Frontmatter

Chapter 1. 2D/3D Image Data Analysis for Object Tracking and Classification

Abstract

Object tracking and classification is of utmost importance for different kinds of applications in computer vision. In this chapter, we analyze 2D/3D image data to address solutions to some aspects of object tracking and classification. We conclude our work with a real time hand based robot control with promising results in a real time application, even under challenging varying lighting conditions.

Seyed Eghbal Ghobadi, Omar Edmond Loepprich, Oliver Lottner, Klaus Hartmann, Wolfgang Weihs, Otmar Loffeld

Chapter 2. Robot Competence Development by Constructive Learning

Abstract

This paper presents a constructive learning approach for developing sensor-motor mapping in autonomous systems. The system’s adaptation to environment changes is discussed and three methods are proposed to deal with long term and short term changes. The proposed constructive learning allows autonomous systems to develop network topology and adjust network parameters. The approach is supported by findings from psychology and neuroscience especially during infants cognitive development at early stages. A growing radial basis function network is introduced as a computational substrate for sensory-motor mapping learning. Experiments are conducted on a robot eye/hand coordination testbed and results show the incremental development of sensory-motor mapping and its adaptation to changes such as in tool-use.

Q. Meng, M. H. Lee, C. J. Hinde

Chapter 3. Using Digital Watermarking for Securing Next Generation Media Broadcasts

Abstract

The Internet presents a problem for the protection of intellectual property. Those who create content must be adequately compensated for the use of their works. Rights agencies who monitor the use of these works exist in many jurisdictions. In the traditional broadcast environment this monitoring is a difficult task. With Internet Protocol Television (IPTV) and Next Generation Networks (NGN) this situation is further complicated.

In this work we focus on Digitally Watermarking next generation media broadcasts. We present a framework which provides the ability to monitor media broadcasts that also utilises a Public Key Infrastructure (PKI) and Digital Certificates. Furthermore, the concept of an independent monitoring agency, that would operate the framework and act as an arbiter, is introduced. We finally evaluate appropriate short signature schemes, suitable Watermarking algorithms and Watermark robustness.

Dominik Birk, Seán Gaines

Chapter 4. A Reduced-Dimension Processor Model

Incorporating Microarchitectural Parameters and Software’s Dynamic Characteristics

Abstract

Architectural simulators used for microprocessor design study and optimization can require large amount of computational time and/or resources. In such cases, models can be a fast alternative to lengthy simulations, and can help reach a designer near-optimal system configuration. However, the non-linear characteristics of a processor system make the modeling task quite challenging. The models not only need to incorporate the micro-architectural parameters but also the dynamic behavior of programs. This paper presents a hybrid (hardware/software), non-linear model for processors. The model provides accurate predictions of processor throughput for a wide range of design space. We used different groups of code basic blocks to investigate their relationships to the execution efficiency of a superscalar processor. For this purpose, we utilized the frequencies of the blocks to represent runtime nature of ten benchmark programs. We were able to reduce the number of hardware and software parameters by employing correlation coefficients and principal component analysis.

Azam Beg

Chapter 5. Hybrid Machine Learning Model for Continuous Microarray Time Series

Abstract

A hybrid machine learning model of the principal component analysis and neural network is described for the continuous microarray gene expression time series. The methodology can model numerically the continuous gene expression time series. The proposed model can give us the extracted features from the gene expressions time series with higher prediction accuracies. It can help practitioners to gain a better understanding of a cell cycle, and to find the dependency of genes, which is useful for drug discoveries. In this chapter, we describe the background, the machine learning algorithms, and then the application of the hybrid machine learning in the microarray analysis. The machine learning model is compared with other popular continuous prediction methods. Based on the results of two public microarray datasets, the hybrid method outperforms the other continuous prediction methods.

Sio-Iong Ao

Chapter 6. An Asymptotic Method to a Financial Optimization Problem

Abstract

This paper studies the borrower’s optimal strategy to close the mortgage when the volatility of the market investment return is small. Integral equation representation of the mortgage contract value is derived, then used to find the numerical solution of the free boundary. The asymptotic expansions of the free boundary are derived for both small time and large time. Based on these asymptotic expansions two simple analytical approximation formulas are proposed. Numerical experiments show that the approximation formulas are accurate enough from practitioner’s point of view.

Dejun Xie, David Edwards, Giberto Schleiniger

Chapter 7. Analytical Design of Robust Multi-loop PI Controller for Multi-time Delay Processes

Abstract

In this chapter, a robust design of multi-loop PI controller for multivariable processes in the presence of the multiplicative input uncertainty is presented. The method consists of two major steps: firstly, the analytical tuning rules of multi-loop PI controller are derived based on the direct synthesis and IMC-PID approach. Then, in the second step, the robust stability analysis is utilized for enhancing the robustness of proposed PI control systems. The most important feature of the proposed method is that the tradeoff between the robust stability and performance can be established by adjusting only one design parameter (i.e., the closed-loop time constant) via structured singular value synthesis. To verify the superiority of the proposed method, simulation studies have been conducted on a variety of the nominal processes and their plant-model mismatch cases. The results demonstrate that the proposed design method guarantees the robustness under the perturbation on each of the process parameters simultaneously.

Truong Nguyen Luan Vu, Moonyong Lee

Chapter 8. Automatic and Semi-automatic Methods for the Detection of Quasars in Sky Surveys

Abstract

With the advances of the technologies for the sky surveys, massive amount of survey data become available. It would be very helpful for the automatic and semi-automatic methods in the classifications/detections of the astrophysical objects. In fact, for surveys of millions of objects, it may not be possible to detect the desired objects by expert inspection alone. Quasars are interesting astrophysical objects that have been recently discovered more comprehensively from the sky surveys. Automatic and semi-automatic methods have been proposed for the detection of the quasars from the massive data produced by the modern sky surveys. In this chapter, the first section describes about the existing automatic and semi-automatic methods for the comprehensive search of quasars. Secondly, it will be explored to see if some machine learning algorithms can automatically classify the light curves of the quasars against the very similar light curves of the other stars. For example in MACHO sky survey, the light curves of the Be stars are so similar with the quasar light curves that the previous algorithms and even the manual examination by experts cannot tell the difference between the light. Experimental results will also be shown for this exploratory work.

Sio-Iong Ao

Chapter 9. Improving Low-Cost Sail Simulator Results by Artificial Neural Networks Models

Abstract

In the present study a method is proposed to reduce the error level of these simplified simulators by correcting the results achieved by means of neural network based approximations. The results of simple aerodynamic simulators used within an evolutionary sail design process are used as application example. The neural network correction is carried out in this case by comparing the numerical results with wind tunnel experiments performed on sail models.

V. Díaz Casás, P. Porca Belío, F. López Peña, R. J. Duro

Chapter 10. Rough Set Approaches to Unsupervised Neural Network Based Pattern Classifier

Abstract

Unsupervised neural network based pattern classification is a widely popular choice for many real time applications. Such applications always face challenges of processing data with lot of consistency, inconsistency, ambiguity or incompleteness. Hence to deal with such challenges a strong approximation tool is always needed. Rough set is one such tool and various approaches based on Rough set, if are applied to pure neural (unsupervised) pattern classifier can yield desired results like faster convergence, feature space reduction and improved classification accuracy. The application of such approaches at respective level of implementation of neural network based pattern classifier for two case studies are discussed here. Whereas more emphasis is given on the preprocessing level based approach used for feature space reduction.

Ashwin Kothari, Avinash Keskar

Chapter 11. A New Robust Combined Method for Auto Exposure and Auto White-Balance

Abstract

This paper proposes a new auto-exposure and auto white-balance algorithm that can accurately detect high-contrast lighting conditions and improve the dynamic range of output images for a camera system. The proposed method calculates the difference between the mean value and the median value of the brightness level of captured pictures to estimate lighting conditions. After that, a multiple exposure mechanism which can improve image details is carried out in combination with a simple auto white-balance algorithm which is capable of detecting pictures with one primary color. Simulation results show that the system works well with CMOS sensors used in mobile phones and surveillance cameras. Besides, the proposed algorithm is fast and simple and therefore can be fitted in most CMOS platforms that have limited capabilities.

Quoc Kien Vuong, Se-Hwan Yun, Suki Kim

Chapter 12. A Mathematical Analysis Around Capacitive Characteristics of the Current of CSCT: Optimum Utilization of Capacitors of Harmonic Filters

Abstract

A new shunt reactive power compensator, CSCT, is presented and introduced in this paper. Mathematical analysis of harmonic content of the current of CSCT is performed and use of a winding with additional circuit has been presented as a solution to suppress these harmonics.

Mohammad Golkhah, Mohammad Tavakoli Bina

Chapter 13. Harmonic Analysis and Optimum Allocation of Filters in CSCT

Abstract

Mohammad Golkhah, Mohammad Tavakoli Bina

Chapter 14. Digital Pen and Paper Technology as a Means of Classroom Administration Relief

Mobile Tools for Teachers (MTT) Evaluation Study

Abstract

This paper contains the results of the Mobile Tools for Teachers project concerning the viability of digital pen and paper technology (DPPT) for administration in a K-12 classroom environment. Filled out forms were evaluated and interviews as well as user tests with teachers were done to show the advantages and disadvantages of DPPT compared to regular methods for attendance tracking and grading. Additionally, the paper addresses the problems that arise with DPPT in a classroom environment and includes suggestions how to deal with those.

Jan Broer, Tim Wendisch, Nina Willms

Chapter 15. A Conceptual Model for a Network-Based Assessment Security System

The Virtual Invigilator

Abstract

The use of computer to assess learning is increasing at colleges and university as the use of technology on campuses increase. The challenge for the instructors at these institutions is to find a way to ensure the integrity of the assessments while still allow students to access network resources during the assessment. A variety of approaches exist that attempt to create an electronic environment that allows students to access only the resources that are permitted. Unfortunately it is nearly impossible to build a system that allow access to the set of resources that a instructor chooses while guaranteeing that no other resources is being accessed. This paper provides an alternate approach to the challenge of securing an assessment and presents a model of a system that can be used to ensure the integrity of the assessment even when unrestricted access to the network is provided.

Nathan Percival, Jennifer Percival, Clemens Martin

Chapter 16. Incorrect Weighting of Absolute Performance in Self-Assessment

Abstract

Students spend much of their life in an attempt to assess their aptitude for numerous tasks. For example, they expend a great deal of effort to determine their academic standing given a distribution of grades. This research finds that students use their absolute performance, or percentage correct as a yardstick for their self-assessment, even when relative standing is much more informative. An experiment shows that this reliance on absolute performance for self-evaluation causes a misallocation of time and financial resources. Reasons for this inappropriate responsiveness to absolute performance are explored.

Scott A. Jeffrey, Brian Cozzarin

Titel: Advances in Machine Learning and Data Analysis
herausgegeben von: Mahyar A. Amouzegar
Verlag: Springer Netherlands
Electronic ISBN: 978-90-481-3177-8
Print ISBN: 978-90-481-3176-1
DOI: https://doi.org/10.1007/978-90-481-3177-8