B.J. van Wyk

A RKHS interpolator-based graph matching algorithm

We present an algorithm for performing attributed graph matching. This algorithm is derived from a generalized framework for describing functionally expanded interpolators which is based on the theory of reproducing kernel Hilbert spaces (RKHS). The algorithm incorporates a general approach to a wide class of graph matching problems based on attributed graphs, allowing the structure of the graphs to be based on multiple sets of attributes. No assumption is made about the adjacency structure of the graphs to be matched.

A LEARNING-BASED FRAMEWORK FOR GRAPH MATCHING

This paper presents a unifying review of a learning-based framework for kernel-based attributed graph matching. The framework, which includes as special cases the RKHS Interplator-Based Graph Matching (RIGM) and Interpolator-Based Kronecker Product Graph Matching (IBKPGM) algorithms, incorporates a general approach where no assumption is made about the adjacency structure of the graphs to be matched. Corresponding pairs of attributed adjacency matrices and attribute vectors of an input and reference graph are used as the input–output training set of a constrained multi-input multi-output multi-variable mapping to be learned. It is shown that a Reproducing Kernel Hilbert Space (RKHS) based interpolator can be used to infer this mapping. Partially constraining the inferred mapping by the generation of additional consistency input–output training pairs and the use of polynomial feature augmentation lead to improved performance. The proposed learning-based framework avoids the explicit calculation of compatibility values.

An interpolator-based network for graph matching

An interpolator-based network for image matching is presented. Network weights are determined using a generalized framework for describing functionally expanded interpolators which is based on the theory of reproducing kernel Hilbert spaces (RKHS). The algorithm implicitly assumes that images can be represented using attributed undirected relational graphs, thereby reducing the matching problem to the inference of a permutation sub-matrix from the interpolator-based network. The algorithm incorporates a general approach to a wide class of graph-based matching problems.An interpolator-based network for image matching is presented. Network weights are determined using a generalized framework for describing functionally expanded interpolators which is based on the theory of reproducing kernel Hilbert spaces (RKHS). The algorithm implicitly assumes that images can be represented using attributed undirected relational graphs, thereby reducing the matching problem to the inference of a permutation sub-matrix from the interpolator-based network. The algorithm incorporates a general approach to a wide class of graph-based matching problems.

Teaching protocol exchanges over cellular air interfaces

The evolutionary path taken by cellular standards to the current and future standards is incomplete without fully understanding the older standards. The comprehension of the GSM standard, specifically the procedures for protocols exchange over the air interface will help students understand radio resource allocation procedures in GPRS and UMTS, and will ultimately assist future communication engineers to be able to design and solve problems related to these cellular standards. In this paper we describe the novel architecture of our teaching software, developed and programmed in Java. We present all the user interfaces and show how it can be used to assist in teaching, by presenting how the mobile station and the GSM/GPRS network exchange data frames over the air interface with their associated protocols. Finally the uniqueness of this teaching tool was proved from the stand point of our modular architecture.

Intelligent network planning of cellular networks for Southern Africa

The rapid growth of telecommunication networks worldwide has placed greater demands and challenges on conventional radio planning tools. One way of overcoming some of these demands is to introduce intelligent algorithms that are capable of extracting valuable knowledge from existing data to support the planning process. Various forms of intelligent methods could be utilized for this purpose. An overview of the use of various intelligent algorithms that have been proposed for the purposes of optimizing the radio planning process is provided and considers an appropriate method that could be used for planning networks effectively for the South African context

DOTTA: a dynamic VLIW like microprocessor framework for high speed parallel computing

In this paper, a new microprocessor design framework, called DOTTA (dynamic operation transport triggered array) is introduced. An FPGA (field programmable gate array) implementation of DOTTA is presented. The aim of this new processor framework is to eliminate bottlenecks introduced by traditional microprocessor architectures. The framework defines a task-specific microprocessor that is application customisable on the target system it is operating on; this has been achieved by using Xilinxtrade FPGA fabric

Successive projection aerial image matching

A successive projection algorithm for matching serial images is presented. The algorithm is based on an approximate projections onto convex sets (POCS) method and uses a graph-based technique in its formulation. It is unique in the way a constrained cost function is minimized using POCS methodology. Simulation results indicate that our algorithm compares favourably to other well-known graph-based algorithms. The suitability of our algorithm for matching infrared rooftop images is highlighted.