• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 4132
  • 1269
  • 509
  • 267
  • 200
  • 119
  • 91
  • 46
  • 45
  • 45
  • 45
  • 45
  • 45
  • 43
  • 35
  • Tagged with
  • 8353
  • 2403
  • 1633
  • 1129
  • 1121
  • 1106
  • 1037
  • 973
  • 919
  • 917
  • 846
  • 792
  • 726
  • 551
  • 524
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
41

Multiple control levels in structured genetic alogorithms

Molfetas, Angelos, University of Western Sydney, College of Health and Science, School of Computing and Mathematics January 2006 (has links)
This work examines the impact that genes with activation relationships have on Genetic Algorithms (GAs). These activation relationships allow genes to control whether other associated genes get expressed in the phenotype. More specifically, this thesis investigates the effect that the incorporation of control levels (tiers of genes which determine the activation of lower genes) have on GAs which are used to generate feed forward Artificial Neural Networks (ANNs). In order to evaluate the performance of different levelled Structured Genetic Algorithms (SGAs; they are GAs which possess control levels), numerous experiments were conducted, utilising the four input XOR, Mackey-Glass and Breast Cancer data sets. In addition, the thesis derives four mathematical models which describe how SGA redundancy changes as more control levels are incorporated. This thesis also presented and implemented a novel four level Structurally Evolved Neural Network Algorithm (SENNGA). Furthermore, it demonstrated the novel implementation of a three level SENNGA. Empirical results show that the incorporation of control levels can improve convergence speed and accuracy, up until the optimum number of control levels is reached. In all experiments increasing the number of control levels had the effect of encouraging the production of leaner and more efficient ANNs. Three and four level SENNGAs can demonstrate superior performance in terms of reducing the training error and generalisation. However this does not apply in all cases. Of particular note, three and four level SENNGAs have exhibited a reduced level of generalisation in the breast cancer experiments, compared two level ones. / Doctor of Philosophy (PhD)
42

Table look-up CORDIC : effective rotations through angle partitioning

Arbaugh, Jason, Todd, 1970- 08 August 2011 (has links)
Not available
43

A subsolutions approach to the analysis and implementation of splitting algorithms in rare event simulation.

Dean, Thomas Anthony. January 2008 (has links)
Thesis (Ph.D.)--Brown University, 2008. / Advisor: P. G. Dupuis. Includes bibliographical references (leaves 172-174).
44

Deployable architecture

James, Andre. January 2008 (has links)
Thesis (M. S.)--Architecture, Georgia Institute of Technology, 2008. / Committee Chair: Spuybroek, Lars; Committee Member: Reither, Gernot; Committee Member: Young, Kevin. Part of the SMARTech Electronic Thesis and Dissertation Collection.
45

Theory and computational studies of mechanochemical phenomena

Konda, Sai Sriharsha Manoj Varma 30 June 2014 (has links)
Mechanochemistry, or the modulation of chemical reactivity through the application of mechanical forces, has shown to facilitate a number of otherwise prohibitive chemical transformations. Computational approaches employing electronic structure calculations have explained a number of mechanochemically activated processes such as thermally inaccessible isomerizations and cycloreversions, symmetry-forbidden electrocyclic ring openings or activation of latent catalysts and, more recently, have been successfully used to design novel mechanosensitive systems. A significant limitation of such approaches, however, is their high computational cost, as finding force dependent transition states requires multiple saddle searches and consequently, multiple energy evaluations. To circumvent this problem, an approximation has been proposed, extending the well know "Bell formula", which estimates the force-dependent reaction barrier based on zero-force transition state properties. We demonstrate the numerical efficiency of this approximation termed as extended Bell theory (EBT) by comparing to existing theories and experiments. We also apply this method to suggest the unexplored, yet potentially useful possibility of suppressing chemical reactions through mechanical perturbation. Furthermore, in sharp contrast to simple, one-dimensional theories, our analysis reveals that the anti-Hammond effect is dominant in the mechanical activation of polyatomic molecules. Finally, we propose a numerical scheme to address the drawback of the EBT approximation, which is the failure to account for force-induced instabilities. Our approach provides a computationally efficient recipe to track the instabilities and follow the evolution of the reactant or transition states at any explicit force. We provide a classification of the different instability scenarios, and provide an illustrative example for each case. / text
46

Better algorithms for comparing trees

Fok, Chi-wai., 霍志偉. January 2010 (has links)
published_or_final_version / Computer Science / Master / Master of Philosophy
47

Flexgraph: flexible subgraph search in large graphs

Yuan, Wenjun., 袁文俊. January 2010 (has links)
published_or_final_version / Computer Science / Master / Master of Philosophy
48

Table look-up CORDIC: effective rotations through angle partitioning

Arbaugh, Jason, Todd 28 August 2008 (has links)
Not available / text
49

Adaptive CORDIC: using parallel angle recording to accelerate CORDIC rotations / Using parallel angle recording to accelerate CORDIC rotations

Rodrigues, Terence Keith, 1974- 29 August 2008 (has links)
This dissertation presents the Parallel Angle Recoding algorithm which is used to accelerate the calculation of CORDIC rotations. The CORDIC algorithm is used in the evaluation of a wide variety of elementary functions such as Sin, Cos, Tan, Log, Exp, etc. It is a simple and versatile algorithm, but its characteristic linear convergence causes it to suffer from long latency. It can be sped up by using the angle recoding algorithm which skips over certain intermediate CORDIC iterations to deliver the same precision while requiring 50% or fewer iterations. However because the selection of the angle constants is quite complex and must be performed off-line, its use has been limited to applications where the rotation angle is static and known a priori. This dissertation extends the low-latency advantage of the angle recoding method to dynamic situations too, where the incoming angle of rotation is allowed to take on any arbitrary value. The proposed method is called Parallel Angle Recoding and it makes use of a much simpler angle selection scheme to identify the angle constants needed by angle recoding. Because of its simplicity, it can be easily implemented in hardware without having to increase the cycle time. All the angle constants for angle recoding can be found in parallel in a single preliminary step by testing just the initial incoming rotation angle using range comparators -- there is no need to perform successive CORDIC iterations in order to identify them. With increasing precision, (N= 8, 16, 24, 32, etc.) the number of comparators which are needed by this scheme increases rapidly. The parallel angle recoding method can be re-formulated to apply to smaller groups of consecutive angle constants known as 'sections.' This limits the number of comparators that are needed, to a reasonable amount. There is an attendant savings in area and power consumption, but at the same time the evaluation of multiple sections introduces additional overhead cycles which reduces some of the gains made in latency by the Parallel Angle Recoding method. By interleaving multiple rotations and making use of a small buffer to store intermediate results, the number of overhead cycles can be reduced drastically. The Parallel Angle Recoding technique is modelled using Verilog, synthesised and mapped to a 65 nm. cell library. The latency and area characteristics that are obtained show that the method can improve the performance of the rotation mode in CORDIC, by delivering a reduced iteration count with no increase in the cycle time, and only a modest increase in power and area.
50

Overlapping clustering

Krumpelman, Chase Serhur 13 December 2010 (has links)
Analysis of large collections of data has become inescapable in many areas of scientific and commercial endeavor. As the size and dimensionality of these collections exceed the pattern recognition capability of the human mind computational analysis tools become a necessity for interpretation. Clustering algorithms, which aim to find interesting groupings within collections of data, are one such tool. Each algorithm incorporates into its design an inherent definition of “interesting” intended to capture nonrandom data groupings likely to have some interpretation to human users. Most existing algorithms include as part of their definition of “interesting” an assumption that each data point can belong at most to one grouping. While this assumption allows for algorithmic convenience and ease of analysis, it is often an artificial imposition on true underlying data structure. The idea of allowing points to belong to multiple groupings - known as “overlapping” or “multiple membership” clustering - has emerged in several domains in ad hoc solutions lacking conceptual unity in approach, interpretation, and analysis. This dissertation proposes general, domain-independent elucidations and practical techniques which address each of these. We begin by positing overlapping clustering’s role specifically, and clustering’s role in general, as assistive technologic tools allowing human minds to represent and interpret structures in data beyond the capability of our innate senses. With this guiding purpose clarified, we provide a catalog of existing techniques. We then address the issue of objectively comparing the results of different algorithms, specifically examining the previously defined Omega index, as well as multiple membership generalizations of normalized mutual information. Following that comparison, we propose a novel approach to com- paring clusterings called cluster alignment. By combining a sorting algorithm with a greedy matching algorithm, we produce comparably organized membership matrices and a means for both numerically and visually comparing multiple-membership assignments. With overlapping clustering’s purpose defined, and the means to analyze results, we move on to presenting algorithms for efficiently discovering overlapping clusters in data. First, we present a generalization of one of the common themes in the ad hoc approaches: additive clustering. Starting with a previously developed structural model of additive clustering, we generalize it to be applicable to any regular exponential family distribution thereby extending its utility into several domains, notably high-dimensional sparse domains including text and recommender systems. Finally, we address overlapping clustering by examining the properties of data in similarity spaces. We develop a probabilistic generative model of overlapping data in similarity spaces, and then develop two conceptual approaches to discovering overlapping clustering in similarity spaces. The first of these is the conceptual multiple-membership generalization of hierarchical agglomerative clustering, and the second is an iterative density hill-climbing algorithm. / text

Page generated in 0.0508 seconds