Completed MS Projects/2006-2007
From CSWiki
Contents |
[edit] Geetha Akula
Advisor: V. Crespi
Disordered networks, such as small-world networks are the focus of recent interest because of their potential as models for the interaction networks of complex systems. Specifically, neither random networks nor regular lattices seem to be an adequate framework within which to study ‘real-world’ complex systems such as chemical-reaction networks, neuronal networks, food webs, social networks, scientific collaboration networks, and computer networks. Small-world networks, which emerge as the result of randomly replacing a fraction P of the links of a d dimensional lattice with new random links, interpolate between the two limiting cases of a regular lattice (P equals to 0) and a random graph (P equeals to 1). A small-world network is characterized by the following properties: (i) the local neighborhood is preserved (as for regular lattices); and (ii) the diameter of the network, quantified by average shortest distance between two vertices, increases logarithmically with the number of vertices n (as for random graphs). The latter property gives the name small-world to these networks, because it is possible to connect any two vertices in the network through just a few links, and the local connectivity would suggest the network to be of finite dimensionality. The structure of small-world networks and of real networks has been probed through the calculation of their diameter as a function of network size. In particular, networks such as the electric power grid for Southern California, the network of movie-actor collaborations, the neuronal network of the worm Caenorhabditis elegans seem to be small-world networks. These three networks as well as the world-wide web are scale-free – that is, they have a distribution of connectivities that decays with a power law tail. Scale-free networks emerge in the context of a growing network in which new vertices connect preferentially to the more highly connected vertices in the network. Scale-free networks are also small-world networks, because (i) they have clustering coefficients much larger than random networks and (ii) their diameter increases logarithmically with the number of vertices n. The thesis focuses on the study of various aspects of small world networks. The explanation of the phenomenon, properties, examples of various networks that exhibit small world properties.
The presentation on SMALL WORLDS.
[edit] Ruben Campos
Model Based Programming: Executable UML with Sequence Diagrams
Advisor: Dr. Russell Abbott
The history of programming can be described in terms of how the level of abstraction has been raised with each new generation of programming. Whether its going from Assembly Language to the C programming language or going from Procedural languages to Object-Oriented programming, the transition between programming generations always seems to be defined by the higher level of abstraction that was achieved. With the OMG pushing the Model Driven Architecture using OMG standards such as the UML, the next logical generational shift is a transition from Object-Oriented Programming to Model Based Programming. The concept of Executable UML appears to provide the most potential to become the core for Model Based Programming. Although an official Executable UML Specification is still a work in progress, there have been attempts to define Executable UML. One such attempt is by Mellor and Balcer, who have designed a rendition of Executable UML albeit with design features that would not be compatible with true model based programming. This paper describes how an alternative Executable UML design can achieve true model based programming.
The Presentation on Model Based Programming: Executable UML with Sequence Diagrams.
[edit] ZhengHui Hu
An Automatic License Plate Recognition System using Image Processing and Neural Network
Advisor: Dr. EunYoung Kang
Many challenges are involved in developing a reliable and efficient automatic License Plate Recognition System (LPRS). Being able to cope with images taken under different conditions, where lighting, size and location of the plate change has been a topic of many researches in computer vision and pattern recognition.
In this Thesis, innovative methods are proposed for license plate recognition that are targeted to solve the inherited issues. For plate localization, a heuristic combining several traditional image processing technique is used. Techniques such as histogram equalization, edge detection, filtering and component analysis each plays a role in the extraction process. For character segmentation, a interpolation algorithm based on information obtained by filtering and adaptive thresholding is used to separate each character. Artificial neural networks (ANN) are in charge of the Optical Character Recognition. Traditional training method for ANN are extremely time consuming and often result in sub-optimal configurations. A hybrid training method is introduced by combining traditional gradient descent based Back-propagation and random selection based Simulated Annealing process to overcome such shortcomings.
The presentation on An Automatic License Plate Recognition System using Image Processing and Neural Network, March 19, 2007.
[edit] Alla Lanovenko
Dynamic Group Key Management in Outsourced Databases
Advisor: H. Guo
Database outsourcing is becoming increasingly popular introducing a new paradigm, called database-as-a-service, where an organization’s database is stored at an external service provider. In such a scenario, access control is a very important issue, especially if the data owner wishes to publish his or her data for external use. Previous researchers had done many efforts for designing and querying encrypted outsourced databases focusing on the challenges posed by protecting data at the server side. However, protecting outsourced database from the unauthorized access on the client side is still an open issue. Since, knowledge of the decryption key allowed clients to not only access authorized data, but also entire outsourced database, which violate data confidentiality and owners privacy. This thesis addresses the problem of the access control on the client side. It addresses one of the major deficiencies of the outsourced database model: confidentiality. The thesis proposes the key management schema suitable for the dynamic environment. This schema can be efficiently applied to the outsourced databases and is based on the widely used Rivest-Shamir-Adelman (RSA) cryptographic algorithm.
The presentation on Dynamic Group Key Management in Outsourced Databases.
[edit] Yuehong Liao
A Java Instrumentation-Based Approach For the Performance Management of J2EE Applications
Advisor: Dr. Raj Pamula
J2EE has been widely used in the software system development at enterprise level, so the analysis of the performance and related dynamic behavior of J2EE applications are becoming more and more important to the software development. This thesis presents a novel approach to collect runtime information of the J2EE applications and help developers to analyze the dynamic behaviors and unearth the bottlenecks based on the Java instrumentation. This Java instrumentation approach is to insert the probe codes into the class byte codes when the classes are loaded into the JVM space by changing the standard class loader according to developers’ specified Java instrumentation task. A software tool – JPManager – a J2EE performance management system is also discussed in this thesis.
[edit] Andre Liv
Development of Web-based SimEconInstructional Software
Advisor: C. Sun
Abstract: With the popularity of the World Wide Web, more and more applications are becoming web-based because of its flexible and dynamic contents. Web-based applications allow users to interact with the applications with much more responsiveness. My project will be converting SimEcon modules to web-based format. SimEcon is an instructional tool designed to help students learn about Principles of Economics, add value to Economics education, and increase instructional access. SimEcon consists of sixteen modules which include Markets, Competition, Monopoly, Labor, Macro1, DrugWars, Banking, E-Growth, UtilityMax, CostMin, GasStations, Externalities, Macro2, Macro3, Exchange$, and $Distribution. For this project, eight modules will be converted to web-based format using ASP.NET written with VisualBasic.NET.
The Presentation on Development of Web-based SimEconInstructional Software.
[edit] Roudabeh Moraghebi
Advisor: Dr. Jiang Guo
This thesis describe the development and initial experiences with a web base system, Tiger E-learning management system developed and designed to facilitate teachers in the management of educational course for their students, especially by helping teachers and students with course administration.The most important goal includes providing lecture materials, resources, increasing transparency, getting fast feedback and increasing contact with and among students.
[edit] Martin J. Olsen
An Eclipse-Based Web Service Composition Tool
Advisor: Dr. Jiang Guo
The thesis describes the development of a graphical plug-in for Eclipse in which the user can create a web service composition. By using web service composition, different web services can be connected through their inputs and outputs. The project includes different technologies such as web service composition, Java, Web services, Eclipse plug-in development and more. Eclipse GEF is used as the editing framework. Dynamic, stub-less, invocation serves as basis for the connection framework enabling connection to any kind of web services regardless of design and implementation. The user will be required to specify the location of a WSDL file, either manually or through UDDI upon which the system will build a local model of the web service structure. The plug-in is given the name “Eclipse Web Service Architecture Plug-in”, or just as an abbreviation: “EWAP”. The EWAP plug-in serves as one of the first implemented web service composition tools for Eclipse.
The presentation on the Eclipse-Based Web Service Composition tool.
[edit] Snehal Patel
Peer Discovery Protocol For JXTA Peer-to-Peer Network
Advisor: H. Guo
Peer-to-peer (P2P) networking has recently emerged as a new paradigm for resource sharing, for building distributed network, and for grid technology. Peer-to-peer networks offer the significant advantages of decentralized, fault-tolerant, and load-balancing systems, with simplicity and ease of use. Despite the multidimensional advantages of Peer-to-peer networks, they face some serious challenges such as flooding, security, and scalability. JXTA is relative newcomer in P2P technologies. JXTA peer discovery mechanism is fully decentralized and it works with query request and query response. This thesis addresses one of the major deficiencies of P2P network: scalability. Additionally, the thesis proposes a schema for peer discovery for JXTA P2P network. The proposed schema can be used for any peer-to-peer networks and is based on a semi-centralized hierarchical tree structure, with edge peers that are grouped together based on their network proximity. Each group will be joined to a service provider called an “agent” and these agents can send requests and respond to messages on behalf of peers. The rendezvous peers are designed as an agent. This type of schema would provide improved scalability than the fully decentralized peer discovery protocol currently incorporated in JXTA.
The presentation on Peer Discovery Protocol was given on 1st December 2006.
[edit] Richard Strom
Hebbian Learning in Multilayer Neural Networks
Advisor: R. Abbott
In this thesis, a Hebbian-based weight adjustment algorithm for multilayer artificial neural networks is presented, and the results of several experimental training sessions using this algorithm are analyzed in order to determine the viability of Hebbian learning in networks required to solve simple problems. Both feedforward and recurrent networks and their applications are explored. It is shown that a balance of Hebbian learning rules, regulated in a biologically plausable way, can lead to useful neural networks. First, it is shown that local Hebbian rules alone can, with high probability, guide a sufficiently large neural network to a meaningful state when trained on simple classification problems. Second, the networks are modified to be recurrent, and it is shown that the networks have pattern recognition capabilities similar to that of a Hopfield network.
The presentation on Hebbian Learning in Multilayer Neural Networks, June 8, 2007.
[edit] John Trafecanty
JWS-QoS: Quality of Service Support Infrastructure for Web Services
Advisor: Dr. Jiang Guo
Web services hold great potential in providing a ubiquitous computing environment where software services are no longer reliant on the underlying language implementation and architecture. Thus far, quality of service standards for web services have been poorly addressed [3]. Web service consumers such as e-businesses will be reluctant to rely on web services for their computing needs if QoS is an unknown factor. Although many proposals have been made to address QoS issues, none have gained wide acceptance. We propose JWS-QoS as a way of determining QoS metrics for today’s web services. JWS-QoS is a software QoS testing suite as well as a web services support infrastructure. JWS-QoS is an implementation of web service standards utilizing many open source tools which focuses on testing three main QoS requirements as defined by the World Wide Web Consortium (W3C), Performance, Reliability, and Capacity.
The Presentation on JWS-QoS: Quality of Service Support Infrastructure for Web Services.
