Computer Sciences Commons^™

A Recurring Improvisational Methodology For Change Management In Erp Implementation, Michelle Sieber, Fiona Fui-Hoon Nah

Research Collection School Of Computing and Information Systems

Change has become an ongoing phenomenon and business process re-engineering has become a necessity for survival of organizations as we approach the 21st century. As participants in this ever-changing working environment, we must be innovative in managing change as it affects both our professional and personal lives. One of the innovative ways to manage change in the implementation of an enterprise resource planning (ERP) system is to adopt a comprehensive change management methodology. In this research, we propose the use of the recurring improvisational change methodology for managing such change. This research is also aimed at implementing the improvisational change …

Website Design: The Concepts Of Informational And Computational Equivalence, Keng Siau

Research Collection School Of Computing and Information Systems

“To the end users, the user interface is the system.” This slogan has been widely used in human-computer interaction field to stress the importance of user interface design. In the Internet and electronic commerce area, the website is not only the system, but it also projects an image of the organization in cyberspace. The design of websites is central to businesses as they create the first impression of organizations to visitors. For a virtual organization, the website is also the only medium that visitors rely on to form their impression of the organization. In this paper, we look at theories …

First 20 Precision Among World Wide Web Search Services (Search Engines), H. Vernon Leighton, Jaideep Srivastava

Library Faculty Works

Five search engines, Alta Vista, Excite, Hotbot, Infoseek, and Lycos, are compared for precision on the first twenty results returned for fifteen queries. All searching was done from January 31 to March 12, 1997. Steps have been taken to ensure that bias has not unduly influence the evaluation. Friedmann's randomized block design is used to perform multiple comparisons for significance. Analysis shows that Alta Vista, Excite and Infoseek are the top three services, with their relative rank changing depending on how one interpreted the concept of "relevant."

Citation:

Leighton, H. Vernon and Jaideep Srivastava. "First 20 Precision among World Wide …

Cross Validation And Mlp Architecture Selection, Timothy L. Andersen, Tony R. Martinez

Faculty Publications

The performance of cross validation (CV) based MLP architecture selection is examined using 14 real world problem domains. When testing many different network architectures the results show that CV is only slightly more likely than random to select the optimal network architecture, and that the strategy of using the simplest available network architecture performs better than CV in this case. Experimental evidence suggests several reasons for the poor performance of CV. In addition, three general strategies which lead to significant increase in the performance of CV are proposed. While this paper focuses on using CV to select the optimal MLP …

Extending The Power And Capacity Of Constraint Satisfaction Networks, Tony R. Martinez, Xinchuan Zeng

Faculty Publications

This work focuses on improving the Hopfield network for solving optimization problems. Although much work has been done in this area, the performance of the Hopfield network is still not satisfactory in terms of valid convergence and quality of solutions. We address this issue in this work by combing a new activation function (EBA) and a new relaxation procedure (CR) in order to improve the performance of the Hopfield network. Each of EBA and CR has been individually demonstrated capable of substantially improving the performance. The combined approach has been evaluated through 20,000 simulations based on 200 randomly generated city …

The Little Neuron That Could, Timothy L. Andersen, Tony R. Martinez

Faculty Publications

SLPs (single layer perceptrons) oflen exhibit reasonable generalization performance on many problems of interest. However, due to the well known limitations of SLPs very little effort has been made to improve their performance. This paper proposes a method for improving the performance of SLPs called "wagging" (weight averaging). This method involves training several different SLPs on the same training data, and then averaging their weights to obtain a single SLP. The performance of the wagged SLP is compared with other more complex learning algorithms (bp, c4.5, ibl, MML, etc) on 15 data sets from real world problem domains. Surprisingly, the …

A Neural Model Of Centered Tri-Gram Speech Recognition, Tony R. Martinez, Dan A. Ventura, D. Randall Wilson, Brian Moncur

Faculty Publications

A relaxation network model that includes higher order weight connections is introduced. To demonstrate its utility, the model is applied to the speech recognition domain. Traditional speech recognition systems typically consider only that context preceding the word to be recognized. However, intuition suggests that considering both preceding context as well as following context should improve recognition accuracy. The work described here tests this hypothesis by applying the higher order relaxation network to consider both precedes and follows context in speech recognition. The results demonstrate both the general utility of the higher order relaxation network as well as its improvement over …

The Robustness Of Relaxation Rates In Constraint Satisfaction Networks, Tony R. Martinez, Dan A. Ventura, D. Randall Wilson, Brian Moncur

Faculty Publications

Constraint satisfaction networks contain nodes that receive weighted evidence from external sources and/or other nodes. A relaxation process allows the activation of nodes to affect neighboring nodes, which in turn can affect their neighbors, allowing information to travel through a network. When doing discrete updates (as in a software implementation of a relaxation network), a goal net or goal activation can be computed in response to the net input into a node, and a relaxation rate can then be used to determine how fast the node moves from its current value to its goal value. An open question was whether …

Combining Cross-Validation And Confidence To Measure Fitness, Tony R. Martinez, D. Randall Wilson

Faculty Publications

Neural network and machine learning algorithms often have parameters that must be tuned for good performance on a particular task. Leave-one-out cross-validation (LCV) accuracy is often used to measure the fitness of a set of parameter values. However, small changes in parameters often have no effect on LCV accuracy. Many learning algorithms can measure the confidence of a classification decision, but often confidence alone is an inappropriate measure of fitness. This paper proposes a combined measure of Cross- Validation and Confidence (CVC) for obtaining a continuous measure of fitness for sets of parameters in learning algorithms. This paper also proposes …

Neural Networks And Structured Knowledge: Knowledge Representation And Reasoning, Franz J. Kurfess

Computer Science and Software Engineering

This collection of articles is the first of two parts of a special issue on "Neural Networks and Structured Knowledge." The contributions to the first part shed some light on the issues of knowledge representation and reasoning with neural networks. Their scope ranges from formal models for mapping discrete structures like graphs or logical formulae onto different types of neural networks, to the construction of practical systems for various types of reasoning. In the second part to follow, the emphasis will be on the extraction of knowledge from neural networks, and on applications of neural networks and structured knowledge to …

Dynamic And Interactive 3d Simulation Over The World Wide Web, Rajesh Vennam

Electrical & Computer Engineering Theses & Dissertations

The thesis presents the development of a new and novel three-dimensional simulation tool over the WWW using Virtual Reality Modeling Language (VRML). VRML is a 3D analog to HTML and serves as a simple, multiplatform language for publishing 3D Web pages. In general, most people use VRML for developing three-dimensional models or worlds that provide little or no interaction for the users. This tool combines interaction and dynamic object creation. A thorough study of VRML is done to explore the features of VRML that could be combined to develop methodologies for creation of dynamic and interactive 3D objects over the …

A Family Of Hierarchical Encoding Techniques For Image And Video Communications, Samah A. Senbel

Computer Science Theses & Dissertations

As the demand for image and video transmission and interactive multimedia applications continues to grow, scalable image and video compression that has robust behavior over unreliable channels are of increasing interest. These desktop applications require scalability as a main feature due to its heterogeneous nature, since participants in an interactive multimedia application have different needs and processing power. Also, the encoding and decoding algorithm complexity must be low due to the practical considerations of low-cost low-power receiver terminals. This requires image and video encoding techniques that jointly considers compression, scalability, robustness, and simplicity.

In this dissertation, we present a family …

Newton Parameter Update Algorithm For Recurrent Neural Networks Applied To Adaptive System Identification And Control, Donald Allen Gates

Electrical & Computer Engineering Theses & Dissertations

This paper shows that the combination of a second-order neural network parameter update algorithm and internal network feedback can be effectively used for adaptive, nonlinear, dynamical system identification and control. Adaptive neural identification and control algorithms are typically utilized for real-time applications where the rate of adaptation is often critical. A fast, adaptive network parameter update algorithm is presented.

Simulation results show that this algorithm is capable of quickly identifying and adapting to changes in system parameters, making it feasible to use for real-time control and fault accommodation applications.

Asymptotic Average Redundancy Of Huffman (And Shannon-Fano) Block Codes, Wojciech Szpankowski

Department of Computer Science Technical Reports

No abstract provided.

A Comparison Of Eclectic Learning And Stagger, J. Cory Barker, Jargalsaihan Batsaihan

Faculty Publications

This project compares two machine-learning methods, Stagger and Eclectic on their classification correctness. Both systems were tested with real-world data sets previously used and tested in other machine learning and statistical literature. The Eclectic System performed better than Stagger on every data set.

Existence Theorems For Scheduling To Meet Two Objectives, April M. Rasala

Dartmouth College Undergraduate Theses

We will look at the existence of schedules which are simultaneously near-optimal for two criteria. First, we will present some techniques for proving existence theorems, in a very general setting, for bicriterion scheduling problems. We will then use these techniques to prove existence theorems for a large class of problems. We will consider the relationship between objective functions based on completion time, flow time, lateness and the number of on-time jobs. We will also present negative results first for the problem of simultaneously minimizing the maximum flow time and average weighted flow time and second for minimizing the maximum flow …

Zero-Parity Stabbing Information, Joseph O'Rourke, Irena Pashchenko

Computer Science: Faculty Publications

Everett et al. [EHN96, EHN97] introduced several varieties of stabbing information for the lines determined by pairs of vertices of a simple polygon P, and established their relationships to vertex visibility and other combinatorial data. In the same spirit, we define the “zero-parity (ZP) stabbing information” to be a natural weakening of their “weak stabbing information,” retaining only the distinction among {zero, odd, even > 0} in the number of polygon edges stabbed. Whereas the weak stabbing information’s relation to visibility remains an open problem, we completely settle the analogous questions for zero parity information, with three results: (1) ZP information …

Analysis Of N-Tier Architecture Applied To Distributed-Database Systems, Alexandre G. Valente

Theses and Dissertations

N-tier architecture has been more commonly used as a methodology for developing large database applications. This work evaluates the use of this architecture instead of the classical Client/Server architecture in developing corporate applications based on distributed databases. The comparison between architectures is performed using applications that execute transactions similar to those defined in the Transaction Process Council Type C benchmark (TPC-C). The environment used for development and testing was the AFIT Bimodal Cluster (ABC); a heterogeneous cluster of PCs, running Microsoft Windows NT 4.0 OS. The comparative experimental analysis demonstrated that the N-tier architecture allows more efficient bandwidth utilization between …

Computers, Art And Smart Rooms: A Smart Picture Frame That Senses The Weather And Genetically Evolves Images, Marisa E. Kolodny

Dartmouth College Undergraduate Theses

By using sensors to sense the environment and genetic programming to evolve images, this thesis explores two methods for developing smart pictures that can be integrated with a living space. The system presented senses the weather and indoor conditions, displays current weather and forecast information retrieved from the web, and displays genetically evolved images. Sensing the weather not only provides the user with information they might find useful, but also allows the computer to gain a better understanding of the user which in turn allows the computer to respond more accurately. Genetic programming allows the computer to better respond to …

A Two Dimensional Crystalline Atomic Unit Modular Self-Reconfigurable Robot, Marsette Arthur Vona Iii

Dartmouth College Undergraduate Theses

Self-reconfigurable robots are designed so that they can change their external shape without human intervention. One general way to achieve such functionality is to build a robot composed of multiple, identical unit modules. If the modules are designed so that they can be assembled into rigid structures, and so that individual units within such structures can be relocated within and about the structure, then self-reconfiguration is possible. We propose the Crystalline Atomic unit modular self-reconfigurable robot, where each unit is called an Atom. In two dimensions, an Atom is square. Connectors at the faces of each Atom support structure formation …

An Application Of Word Sense Disambiguation To Information Retrieval, Jason M. Whaley

Dartmouth College Undergraduate Theses

The problems of word sense disambiguation and document indexing for information retrieval have been extensively studied. It has been observed that indexing using disambiguated meanings, rather than word stems, should improve information retrieval results. We present a new corpus-based algorithm for performing word sense disambiguation. The algorithm does not need to train on many senses of each word; it uses instead the probability that certain concepts will occur together. That algorithm is then used to index several corpa of documents. Our indexing algorithm does not generally outperform the traditional stem-based tf.idf model.

The Implementation Of Dassf Otcl Apis, Hongxia Quan

Dartmouth College Undergraduate Theses

As an extension of Tcl, Otcl provides basic functionality for object-oriented programming in scripting language Tcl. We implemented the Otcl APIs for DaSSF (a parallel simulator software written in C++ at Dartmouth College) using Tclcl software package written in University of California at Berkeley. This document discussed the issues involved in the implementation, especially the communications between C++ objects and Otcl objects required by DaSSF and the naming problems.

Ismis: A Military Transportation Decision-Support Framework, Stephen Goodman, Jens G. Pohl

Collaborative Agent Design (CAD) Research Center

This paper discusses the challenges that the Information Revolution poses to transportation planning, execution and training in the US military services. Attention is drawn to the opportunities provided by global connectivity and the manner in which these opportunities can be applied to advantage in an integrated, collaborative, decision-support framework. The characteristics of military deployment operations are discussed in respect to system requirements such as parallel activities, internal representation of information (rather than data), intelligent assistance, and the integration of planning, execution and training functions.

The Integrated Strategic Mobility Interface System (ISMIS) is described as a framework that is designed to …

A Multilevel Secure Workflow Management System, Myong H. Kang, Judith N. Froscher, Amit P. Sheth, Krzysztof J. Kochut, John A. Miller

Kno.e.sis Publications

The Department of Defense (DoD) needs multilevel secure (MLS) workflow management systems to enable globally distributed users and applications to cooperate across classification levels to achieve mission critical goals. An MLS workflow management system that allows a user to program multilevel mission logic, to securely coordinate widely distributed tasks, and to monitor the progress of the workflow across classification levels is required. In this paper, we present a roadmap for implementing MLS workflows and focus on a workflow builder that is a graphical design tool for specifying such workflows.

Parallel Dassf Discrete-Event Simulation Without Shared Memory, James D. Chalfant

Dartmouth College Undergraduate Theses

The Dartmouth implementation of the Scalable Simulation Framework (DaSSF) is a discrete-event simulator used primarily in the simulation of networks. It achieves high performance through parallel processing. DaSSF 1.22 requires shared memory between all processors in order to operate. This limits the number of processors available and the hardware platforms that can exploit parallelism. We are interested in extending parallel DaSSF operation to architectures without shared memory. We explore the requirements of this by implementing parallel DaSSF using MPI as the sole form of interaction between processors. The approaches used to achieve this can be abstracted and applied to the …

Two Algorithms For Performing Multidimensional, Multiprocessor, Out-Of-Core Ffts, Lauren M. Baptist

Dartmouth College Undergraduate Theses

We show two algorithms for computing multidimensional Fast Fourier Transforms (FFTs) on a multiprocessor system with distributed memory when problem sizes are so large that the data do not fit in the memory of the entire system. Instead, data reside on a parallel disk system and are brought into memory in sections. We use the Parallel Disk Model for implementation and analysis. The first method is a straightforward out-of-core variant of a well-known method for in-core, multidimensional FFTs. It performs 1-dimensional FFT computations on each dimension in turn. This method is easy to generalize to any number of dimensions, and …

Improved Computer Detection And Mapping Of Cerebral Oxygenation, David H. Kung

Dartmouth College Undergraduate Theses

Near-infrared (NIR) optical image reconstruction that incorporates blood oxygen level dependant (BOLD) magnetic resonance imaging has the potential to improve both quantifiable measurement of oxygenation and the spatial resolution involved in such mapping. My thesis continues some preliminary work in this area through development of an analytic diffusion parameter estimation algorithm for use with a NIR imaging array and development of a finite element mesh utility to read a priori BOLD images and tag them with property elements for NIR image resolution improvement.

Investigating Measures For Pairwise Document Similarity, Jeffrey D. Isaacs

Dartmouth College Undergraduate Theses

The need for a more effective similarity measure is growing as a result of the astonishing amount of information being placed online. Most existing similarity measures are defined by empirically derived formulas and cannot easily be extended to new applications. We present a pairwise document similarity measure based on Information Theory, and present corpus dependent and independent applications of this measure. When ranked with existing similarity measures over TREC FBIS data, our corpus dependent information theoretic similarity measure ranked first.

Considerations For An Effective Telecommunications-Use Policy, Michael E. Whitman, Anthony M. Townsend, Robert J. Aalberts

Faculty Articles

Recent changes in federal telecommunications legislation have underscored the importance of an up-to-date and effective telecommunications-use policy in business organizations. With the proliferation of the Internet, intranets, and email as commonplace business tools, the potential for misuse and subsequent liability has become an increasing concern. Even though the recent Supreme Court decision struck down the obscenity provisions of the Communications Decency Act (CDA), it left intact legislation that effectively mandates development of a sound telecommunications-use policy. In addition to potential liability for systems misuse, organizations have also had to address issues of individual employee privacy within the new systems.

This …

Fast Out-Of-Core Sorting On Parallel Disk Systems, Matthew D. Pearson

Dartmouth College Undergraduate Theses

This paper discusses our implementation of Rajasekaran's (l,m)-mergesort algorithm (LMM) for sorting on parallel disks. LMM is asymptotically optimal for large problems and has the additional advantage of a low constant in its I/O complexity. Our implementation is written in C using the ViC* I/O API for parallel disk systems. We compare the performance of LMM to that of the C library function qsort on a DEC Alpha server. qsort makes a good benchmark because it is fast and performs comparatively well under demand paging. Since qsort fails when the swap disk fills up, we can only compare these algorithms …