Digital Commons Network^™

Intelligent Structural Operators For The K-Way Graph Partitioning Problem, Gregor Von Laszewski

Northeast Parallel Architecture Center

A parallel genetic algorithm for the graph partitioning problem is presented, which combines general heuristic algorithms with techniques that are described in evolution theory. In the parallel genetic algorithm the selection of a mate is restricted to a local neighborhood. In addition, the parallel genetic algorithm executes an adaptation step after an individual is generated, with the genetic operators crossover and mutation. During the adaptation step the solution is improved by a common algorithm. Another selection step decides if the adapted descendant should replace the parent individual. Instead of using a uniform crossover operator a more intelligent crossover operator, which …

Optimal Processor Assignment For Pipeline Computations, David Nicol, Rahul Simha, Alok N. Choudhury, Bhagirath Narahari

Electrical Engineering and Computer Science - All Scholarship

The availability of large-scale multitasked parallel architectures introduces the following processor assignment problem for pipelined computations. Given a set of tasks and their precedence constraints, along with their experimentally determined individual response times for different processor sizes, find an assignment of processors to tasks. Two objectives interest us: minimal response given a throughput requirement, and maximal throughput given a response time requirement. These assignment problems differ considerably from the classical mapping problem in which several tasks share a processor; instead, we assume that a large number of processors are to be assigned to a relatively small number of tasks. In …

Pattern Recognition For Nondestructive Evaluation, S. Morris, P. O'Rorke, William E. Bond, M. M. Amirfathi, Daniel C. St. Clair

Computer Science Faculty Research & Creative Works

The issues involved in automating nondestructive evaluation (NDE) techniques are outlined. Attention is given to research focused on the application of machine learning techniques to the construction and maintenance of knowledge-based systems which are capable of evaluating the readings from nondestructive tests that have been performed on aircraft components. Preliminary results obtained from this research are described. In particular, the authors discuss the application of a symbolic machine learning algorithm, ID3, to the NDE problem. ID3 has been used by Douglas Aircraft to classify defects in sets of standard NDE reference blocks. Based on the preliminary results, a need for …

A Visualization Model For Massively Parallel Algorithms, Rashi Khanna, Bruce M. Mcmillin

Computer Science Faculty Research & Creative Works

A visualization model has been developed to analyze the performance of a massively parallel algorithm. Most visualization tools that have been developed so far for performance analysis are based generally on individual processor information and communication patterns. These tools, however, are inadequate for massively parallel computations. It is difficult to comprehend the visual information for many processors. The model, SMILI (Scientific visualization in Multicomputing for Interpretation of Large amounts of Information), addresses this problem by using abstract representations to attain a composite picture which gives better insight to the behavior of the algorithm. Chernoff s Faces have been selected to …

Fault-Tolerant Parallel Matrix Multiplication With One Iteration Fault Detection Latency, Chul Eui Hong, Bruce M. Mcmillin

Computer Science Faculty Research & Creative Works

The checksum technique is a low-cost method to detect errors in matrix operations performed by processor arrays. The fault detection of this method is done only at problem termination, so this method is not an effective fault tolerance technique for large scale matrix multiplication. This paper presents a new algorithm, the ID algorithm, which minimizes the fault-detection latency, In the ID algorithm, a fault is detected as soon as the fault occurs instead of at problem termination. For n2 processors, the fault-latency time of the ID algorithm is l/n of that of checksum algorithm with a run-time penalty of O(nlog2n) …

A Model-Based Approach For Organizing Quantitative Computations, J. Sticklen, A. Kamel, William E. Bond

Computer Science Faculty Research & Creative Works

Model based reasoning (MBR) is currently receiving wide spread attention because it offers a way to circumvent the brittleness of reasoning systems built solely on associational knowledge. To date, most MBR approaches have focused on the use and manipulation of qualitative models. The authors report their experience in applying techniques of functional reasoning to the general problem of organizing quantitative calculations. As a testbed, they have solved a problem initially posed at the Model-Based Diagnosis workshop (Paris, July, 1989): representing an automotive cruise control system. The results show that the principles of the functional reasoning approach can provide leverage in …

Functional Reasoning And Functional Modelling, J. Sticklen, William E. Bond

Computer Science Faculty Research & Creative Works

A car that will not start on a cold winter day and one that will not start on a hot summer day usually indicate two very different situations. When pressed to explain the difference, we would give a winter account- "Oil is more viscous in cold conditions, and that causes . . .'' -and a summer story- "Vapor lock is a possibility in hot weather and is usually caused by . . .'' How do we build such explanations? One possibility is that understanding how the car works as a device gives us a basis for generating the explanations. But …

A Mergeable Double-Ended Priority Queue, S. Olariu, Z. Wen

Computer Science Faculty Publications

An implementation of a double-ended priority queue is discussed. This data structure referred to as min–max–pair heap can be built in linear time; the operations Delete-min, Delete-max and Insert take O(log n) time, while Find-min and Find-max run in O(1) time. In contrast to the min-max heaps, it is shown that two min–max–pair heaps can be merged in sublinear time. More precisely, two min–max–pair heaps of sizes n and k can be merged in time O(log (n/k) * log k).

On A Unique Tree Representation For P4-Extendible Graphs, B. Jamison, S. Olariu

Computer Science Faculty Publications

Several practical applications in computer science and computational linguistics suggest the study of graphs that are unlikely to have more than a few induced paths of length three. These applications have motivated the notion of a cograph, defined by the very strong restriction that no vertex may belong to an induced path of length three. The class of P₄-extendible graphs that we introduce in this paper relaxes this restriction, and in fact properly contains the class of cographs, while still featuring the remarkable property of admitting a unique tree representation. Just as in the case of cographs, the …

Simulating Adaptive Load Sharing Policies On An Ipsc/2 Multicomputer, Yuh Jong Hu, Billy E. Gillett

Computer Science Faculty Research & Creative Works

Unlike most other adaptive load sharing (LS) policy studies, each node in the distributed system is modeled as a central server model represented by a closed queueing network (QN). The primary objective of this study is to use a simulation model to find the improvement for an adaptive LS policy in a distributed system. In homogeneous distributed systems, the simulation results in this study show that the performance improvements between no LS, LS with task placement, and LS with task migration are very small. These results are quite different from other studies, which show a significant improvement of mean response …

Fuzzy Neural Logic Network And Its Learning Algorithms, Fiona Fui-Hoon Nah, Nah Fiona

Research Collection School Of Computing and Information Systems

The paper introduces the basic features of fuzzy neural logic network. Each fuzzy neural logic network model is trained from a set of knowledge in the form of examples using one of the three learning algorithms introduced. These three learning algorithms are the delta rule controlled learning algorithm and two mathematical construction algorithms, namely, the local learning method and the global learning method. Once the fuzzy neural logic network model is constructed, it is ready to accept any unknown input from the user. With a low percentage of mismatched features, output solution can be obtained.

Exclusive And Inclusive Semileptonic Decays Of B Mesons To D Mesons, R. Fulton, M. Thulasidas

Research Collection School Of Computing and Information Systems

We report new measurements of the branching fractions B ( B − → D 0 l − ¯ ν ) , B ( ¯ B 0 → D + l − ¯ ν ) , and B ( B − → D * 0 l − ¯ ν ) . Combining these results with our previous measurement of B ( ¯ B 0 → D * + l − ¯ ν ) , we find that the ratio of semileptonic widths for final states with a vector meson and pseudoscalar meson is ( 2.6 + 1.1 + 1.0 − 0.6 …

Reducing Computational Expense Of Ray-Tracing Using Surface Oriented Pre-Computation, Robert E. Rinker

UNF Graduate Theses and Dissertations

The technique of rendering a scene using the method of ray-tracing is known to produce excellent graphic quality, but is also generally computationally expensive. Most of this computation involves determining intersections between objects in the scene and ray projections. Previous work to reduce this expense has been directed towards ray oriented optimization techniques. This paper presents a different approach, one that bases pre-computation on the characteristics of the scene itself, making the results independent of the position of the observer. This means that the results of one pre-computation run can be applied to renderings of the scene from multiple view …

An Expert Inference Engine For Generation Of Nursing Diagnoses, Tom Edgar

UNF Graduate Theses and Dissertations

Expert computer systems for use in the nursing profession are emerging as a potentially viable alternative to manual procedures. As nursing science continues to develop, the intellectual requirements of assessment and diagnosis are demanding that the professional nurse draw on an ever increasing bank of knowledge to interact effectively with clients. An expert system appears a promising tool to assist the nurse in storing and accessing some of the knowledge necessary to perform the assessment and diagnostic functions.

Problems and opportunities In applying artificial intelligence techniques to nursing science are documented and the current state of expert systems for nursing …

Cascading Of Ever Adaptive Variation Of Lzw Algorithm With Arithmetic Coding Method, Ramesh Vishwanatham, Ketan Tejani

Theses

Two known compression algorithms appropriate for communication are LZW and the Arithmetic Coding method. Both algorithms are adaptive and require no extra communication from the encoder to the decoder. We have considered a variation of LZW algorithm known as Online version which result in better compression. Here we present a scheme to cascade Online LZW with Arithmetic algorithms into an Ever Adaptive data compression algorithm which achieves higher compression ratio and is appropriate for communication.

Three Degrees Of Freedom Compliant Motion Control For Robotic Aircraft Refueling, Clayton M. Andersen

Theses and Dissertations

The Air Force Institute of Technology (AFIT) supports on-going research in application of robotic technology to enhance assembly tasks. As a testbed, AFIT supports the ground-based aerial refueling project, with an emphasis on visual servoing techniques and compliant motion control. This thesis focuses on compliant motion control. Previous research developed the testing environment, which includes a PUMA-560 industrial manipulator, and a half scale mock-up of an aerial refueling system. This effort corrected and expanded the existing two degree of freedom (DOF) compliant controller to three DOF. Three DOF compliance was demonstrated by inserting the refueling nozzle into the receiver port. …

Models And Algorithms For Co-Scheduling Compute-Intensive Tasks On A Network Of Workstations, Mikhail J. Atallah, Christina Lock, Dan C. Marinescu, Howard J. Seigel, Thomas L. Casavant

Department of Computer Science Technical Reports

No abstract provided.

Voronoi Diagrams In A River, Kokichi Sugihara

Department of Computer Science Technical Reports

No abstract provided.

Implementing Expert System Rule Conditions By Neural Networks, Henry Tirri

Department of Computer Science Technical Reports

No abstract provided.

Fitting Semantics For Conditional Term Rewriting, Chilukuri K. Mohan

Electrical Engineering and Computer Science - Technical Reports

This paper investigates the semantics of conditional term rewriting systems with negation which do not satisfy useful properties like termination. It is shown that the approach used by Fitting [5] for Prolog-style logic programs is applicable in this context. A monotone operator is developed, whose fixpoints describe the semantics of conditional rewriting. Several examples illustrate this semantics for non-terminating rewrite systems which could not be easily handled by previous approaches.

The All-Pairs Pipeline, Per Brinch Hansen

Electrical Engineering and Computer Science - Technical Reports

An all-pairs problem is a computation on every possible subset consisting of two elements chosen from a set of n elements. N-body simulation and Householder reduction are all-pairs problems. The paper defines the all-pairs problem concisely by means of precedence matrices and derives a parallel algorithm. The algorithm is presented in both coarse-grain and medium-grain form. The all-pairs paradigm is illustrated by a pipeline for Householder reduction of a matrix to triangular form.

Analyzing Images Containing Multiple Sparse Patterns With Neural Networks, Rangachari Anand, Kishan Mehrotra, Chilukuri K. Mohan, Sanjay Ranka

Electrical Engineering and Computer Science - Technical Reports

We have addressed the problem of analyzing images containing multiple sparse overlapped patterns. This problem arises naturally when analyzing the composition of organic macromolecules using data gathered from their NMR spectra. Using a neural network approach, we have obtained excellent results in using NMR data to analyze the presence of amino acids in protein molecules. We have achieved high correct classification percentages (about 87%) for images containing as many as five substantially distorted overlapping patterns.

A Multi-Scale Approach For Recognizing Complex Annotations In Engineering Documents., Andrew Francis Laine, William Ball, Arun Kumar

All Computer Science and Engineering Research

This paper describes a novel method of character recognition targeted for extracting complex annotations found in engineering documents. The results of this work will make it possible to capture the information contained in documents used to support facilities management and manufacturing. The recognition problem is made difficult in part because characters and text may be expressed in arbitrary fonts and orientations. Our approach includes a novel incremental strategy based on the multi-scale representation of wavelet decompositions. Our approach is motivated by biological mechanisms of the human visual systems. Using wavelets as a set of basis functions, we may decompose an …

Householder Reduction, Per Brinch Hansen

Electrical Engineering and Computer Science - Technical Reports

This tutorial discusses Householder reduction of n linear equations to a triangular form which can be solved by back substitution. The main strengths of the method are its numerical stability and suitability for parallel computing. We explain how Householder reduction can be derived from elementary matrix algebra. The method is illustrated by a numerical example and a Pascal algorithm. We assume that the reader has a general knowledge of vector and matrix algebra but is less familiar with linear transformation of a vector space.

Use Of Grasp Force Focus Positioning To Enhance The Torque Resistance Capability Of Robotic Grasps, Stephen G. Edwards

Theses and Dissertations

Three-point-contact grasps are unique in that the homogeneous solution for the contact forces always produces a grasp force focus. Careful positioning of this focus point in the grasp plane can help avoid two things; slipping at the contact points, and violation of joint torque limits. The focus placement method is explored theoretically by examining two types of grasps on cylinders; 1) fingertip grasps using three independently operated fingers, and 2) single-finger power grasps with one contact point on each of three links. Constraint maps are generated for various fingertip grasps in order to show how proper placement of the grasp …

A Low-Cost Part-Task Flight Training System: An Application Of A Head Mounted Display, David A. Dahn

Theses and Dissertations

To investigate the feasibility of using head-mounted display(HMD) and Computer Image Generators(CIG) in a low cost part-task trainer, we created a prototype system. Our approach was to build a virtual world interface using a HMD to an existing flight simulator application. This allowed a CIG with only one graphics channel to drive the display. To investigate which class of computing platform was suitable for use as the CIG, a cooperative thesis effort was launched to host the simulator on a mini-computer and micro-computer platform. The appropriate CIG could then be determined through demonstration and comparison between the systems. A software …

An Empirical Evaluation Of Analytical Models For Parallel Relational Database Queries, Mark C. Denham

Theses and Dissertations

This thesis documents the design and implementation of three parallel join algorithms to be used in the verification of analytical models developed by Kearns. Kearns developed a set of analytical models for a variety of relational database queries. These models serve as tools for the design of parallel relational database system. Each of Kearns' models is classified as either single step or multiple step. The single step models reflect queries that require only one operation while the multiple step models reflect queries that require multiple operations. Three parallel join algorithms were implemented based upon Kearns' models. Two are based upon …

An Empirical Study Of Combining Communicating Processes In A Parallel Discrete Event Simulation, Ann K. Lee

Theses and Dissertations

The primary goal of distributed discrete event simulations is to achieve speedup in simulation execution time by distributing the processing of the simulation over multiple processors. When partitioned for distribution in this fashion, simulations are typically partitioned such that there are more processes than processors. This thesis reviews existing methods for distributed discrete event simulations, and proposes general guidelines for efficient partitioning for a given communications topology based on empirical evidence. A performance analysis is conducted for two approaches to partitioning the system. The first method chosen is a mapping of multiple processes to a processor and the second approach …

Determining Concurrency In Object-Oriented Design Of Real-Time Embedded Systems Using Ada, Kenneth D. Baum

Theses and Dissertations

One of the characteristics of real-time systems is concurrency. Designers of real-time systems have traditionally determined system concurrency at implementation time using the facilities of a cyclic executive. With the advent of programming language constructs for specifying concurrency, determining concurrency at design time has become a possibility. Several design methods, all of which are extensions of either Structured Design or Jackson System Development, provide heuristics to help the designer make concurrency decisions. The object-oriented approach, however, has no corresponding heuristics to aid designers of real-time systems. The purpose of this thesis was to develop heuristics to help designers make concurrency …

Access And Operator Methods For The Triton Nested Relational Database System, Tina M. Harvey

Theses and Dissertations

Unique database requirements in the realm of non-standard applications (such as computer-aided design (CAD), computer-aided software engineering(CASE), and office information systems (OIS) have driven the development of new data models and database systems based on these new models. In particular, the goal of these new database systems is to exploit the advantages of complex data models that are more efficient (in terms of time and space) than their relationship counterparts. This research effort describes the relational design and implementation of the Triton nested relational database system, a prototype system based on the nested relational data model. Triton is intended to …