Knowledge-based analysis of carotid angiograms

The project described is concerned with the development of a practical system for interpreting carotid angiograms. The general requirements of such a system are identified and some of the design considerations are discussed. The analysis strategies and types of knowledge used by an expert in recognizing and naming vessels and categorizing abnormalities on a single-plane angiogram are described- Two distinct types of knowledge are identified: ‘facts’ knowledge which is drawn from various areas of science, and ‘strategy’ knowledge which determines how the facts should be used. It is suggested that confidence parameters should be associated with each type of knowledge and these should be used in the control of the analysis of an angiogram. The ‘facts’ and ‘strategy’ knowledge bases have a hierarchical structure which, if exploited, would enable the system to be easily adapted for use in other application areas.     

On uncertainty versus size in branching programs

We propose an information-theoretic approach to proving lower bounds on the size of branching programs. The argument is based on Kraft type inequalities for the average amount of uncertainty about (or entropy of) a given input during the various stages of computation. The uncertainty is measured by the average depth of so-called ‘splitting trees’ for sets of inputs reaching particular nodes of the program.

We first demonstrate the approach for read-once branching programs. Then, we introduce a strictly larger class of so-called ‘balanced’ branching programs and, using the suggested approach, prove that some explicit Boolean functions cannot be computed by balanced programs of polynomial size. These lower bounds are new since some explicit functions, which are known to be hard for most previously considered restricted classes of branching programs, can be easily computed by balanced branching programs of polynomial size.     

The architecture of a multi-vector processor system, VPP

The VPP system is a multi-vector processor system which mainly aims at effective satellite image processing. It consists of up to 64 element processors (PUs), an S-D loop network, and an image memory. The PUs can execute flexible vector processing by a new vector access method, ‘Index-set’. The S-D loop network achieves high-speed and contention-free data transfer among the PUs. With these components, a new method for parallel processing, ‘Processor Pipeline’, can be realized on the VPP system.     

Real-time integrated model for visual perception and fuzzy control

Current state-of-the-art security systems incorporate ‘passive’ and/or ‘human’ elements, the effectiveness of which can only be measured by their ability to ‘deter’ intruders. However, rapidly changing economic and cultural conditions have weakened the strengths associated with such systems. In the not too distant future, the need for an ‘active’ security system will become necessary in order to reduce the onslaught of crime.

This paper presents a conceptual basis for the incorporation of artificial intelligence concepts in the design and implementation of ‘active’ security systems. Specifically, the paper discusses issues pertaining to a real-time model for visual perception and tracking of possible intruders.     

On the complexity of ω-type Turing acceptors

Turing machines are considered as recognizers of sets of infinite (ω-type) sequences, so called ω-languages. The basic results on such ω-type Turing acceptors were presented in a preceding paper. This paper focuses on the theory of deterministic ω-type Turing acceptors (ω-DTA's) which turns out to be crucially different from the ‘classical’ theory of Turing machines. It is shown that there exists no ω-DTA which is universal for all ω-DTA's. Two infinite complexity hierarchies for ω-DTA's are established, the ‘states hierarchy’, corresponding to the number of states in the machine, and the ‘designated sets hierarchy’, corresponding to the number of designated sets of states used in the recognition. Concrete examples of ω-languages characterizing each of the complexity classes are exhibited. Two additional examples of interesting ω-languages are presented:

1. (i) An ω-language which is ‘inherently non-deterministic’, i.e. can be recognized by a non-deterministic Turing acceptor but by no deterministic acceptor.

2. (ii) An ω-language which cannot be recognized even by a non-deterministic Turing acceptor.

The above examples are constructed without using diagonalization. Oscillating ω-DTA's, i.e. ω-DTA's which are allowed to oscillate on ω-inputs, are also considered and are shown to be strictly more powerful than non-oscillating ω-DTA's, yet strictly less powerful than non-deterministic ω-Turing acceptors.     

Topological models for boundary representation: a comparison with n-dimensional generalized maps

In boundary representation, a geometric object is represented by the union of a ‘topological’ model, which describes the topology of the modelled object, and an ‘embedding’ model, which describes the embedding of the object, for instance in three-dimensional Euclidean space. In recent years, numerous topological models have been developed for boundary representation, and there have been important developments with respect to dimension and orientability. In the main, two types of topological models can be distinguished. ‘Incidence graphs’ are graphs or hypergraphs, where the nodes generally represent the cells of the modelled subdivision (vertex, edge, face, etc.), and the edges represent the adjacency and incidence relations between these cells. ‘Ordered’ models use a single type of basic element (more or less explicitly defined), on which ‘element functions’ act; the cells of the modelled subdivision are implicitly defined in this type of model. In this paper some of the most representative ordered topological models are compared using the concepts of the n-dimensional generalized map and the n-dimensional map. The main result is that ordered topological models are (roughly speaking) equivalent with respect to the class of objects which can be modelled (i.e. with respect to dimension and orientability).     

Boundary matching algorithm for connected component labelling using linear quadtrees

A computationally fast top-down recursive algorithm for connected component labelling using linear quadtrees is presented. The input data structure used is a linear quadtree representing only black leaf nodes. The boundary matching approach used ensures that at most two adjacencies of any black leaf node are considered. Neighbour searching is carried out within restricted subsets of the input quadtree. The time and space complexities of the algorithm are O(Bn) and O(B) respectively for a linear quadtree with B black leaves constructed from a binary array of size 2ⁿ × 2ⁿ. Simulations show the algorithm to be twice as fast as an existing algorithm that uses an identical input data structure. The top-down algorithm presented can also be used to efficiently generate a linear quadtree representing all nodes — ‘grey’, ‘black’ and ‘white’ — in preorder when given an input linear quadtree representing only ‘black’ leaf nodes. The boundary matching algorithm is computationally fast and has low static and dynamic storage costs, making it useful for applications where linear quadtrees are held in main memory.     

Comparison of several discrete arithmetic schemes for simulating a constrained jet and a lab-scale tangential fired furnace

For tangential fired furnace, the false diffusion will occur when the flow is oblique to the grid lines. It is a major factor to cause serious errors to the prediction of tangential fired furnace. In order to decrease this false diffusion, this study has considered several alternative to the widely used upwind-differencing scheme with the aim of identifying which could be regarded as the most suitable in a general-purpose solving procedure for tangential furnace. The representations adopted are ‘hybrid’, ‘quick’ and ‘27-point’ treatments. Here, ‘27-point’ is a new discrete arithmetic scheme developed by the authors. All these schemes are used to simulate a constrained jet and a lab-scale tangential furnace and to compare solutions with the datum. Overall, the ‘27-point’ approximation emerged as the most satisfactory for simulating tangential furnace.     

Defining a taxonomy of output modalities from an HCI perspective

The paper addresses an issue that must be resolved to produce a scientifically sound and practically useful reference model for intelligent multimedia presentation systems (IMP systems), namely that of providing, from the point of view of Human-Computer Interaction (HCI), a systematic understanding of the types of output information to be presented by IMP systems. The term ‘medium’, as it is used in the context of multimedia systems, is too coarse-grained for distinguishing between different types of output information. The paper introduces the notion of (representational) ‘modalities’ to enable sufficiently fine-grained distinctions to be made. For the term itself to be meaningful, ‘multimodal’ presentations must be composed of unimodal representations. In the approach presented, unimodal representations are defined from a small number of basic properties whose combinations specify the ‘generic’ level of a taxonomy of unimodal output modalities. Additional basic property distinctions serve to generate the more fine-grained ‘atomic’ and ‘sub-atomic’ levels in a hierarchical fashion. The taxonomy is set up with the aim of satisfying four basic requirements, viz. completeness, orthogonality, relevance and intuitiveness. A concluding discussion illustrates the practical use of the taxonomy.     

Getting to know each other—Artificial social intelligence for autonomous robots

This paper proposes a research direction to study the development of ‘artificial social intelligence’ of autonomous robots which should result in ‘individualized robot societies’. The approach is highly inspired by the ‘social intelligence hypothesis’, derived from the investigation of primate societies, suggesting that primate intelligence originally evolved to solve social problems and was only later extended to problems outside the social domain. We suggest that it might be a general principle in the evolution of intelligence, applicable to both natural and artificial systems. Arguments are presented why the investigation of social intelligence for artifacts is not only an interesting research issue for the study of biological principles, but may be a necessary prerequisite for those scenarios in which autonomous robots are integrated into human societies, interacting and communicating both with humans and with each other. As a starting point to study experimentally the development of robots' ‘social relationships’, the investigation of collection and use of body images by means of imitation is proposed. A specific experimental setup which we use to test the theoretical considerations is described. The paper outlines in what kind of applications and for what kind of robot group structures social intelligence might be advantageous.     

Neural networks deterioration models for serviceability condition of buried stormwater pipes

Buried stormwater pipe networks play a key role in surface drainage systems for urban areas of Australia. The pipe networks are designed to convey water from rainfall and surface runoff only and do not transport sewage. The deterioration of stormwater pipes is commonly graded into structural and serviceability condition using CCTV inspection data in order to recognize two different deterioration processes and consequences. This study investigated the application of neural networks modelling (NNM) in predicting serviceability deterioration that is associated with reductions of pipe diameter until a complete blockage. The outcomes of the NNM are predictive serviceability condition for individual pipes, which is essential for planning proactive maintenance programs, and ranking of pipe factors that potentially contribute to the serviceability deterioration. In this study the Bayesian weight estimation using Markov Chain Monte Carlo simulation was used for calibrating the NNM on a case study in order to account for the uncertainty often encountered in NNM's calibration using conventional back-propagation weight estimation. The performance and the ranked factors obtained from the NNM were also compared against a classical model using multiple discrimination analysis (MDA). The results showed that the predictive performance of the NNM using Bayesian weight estimation is better than that of the NNM using conventional backpropagation and MDA model. Furthermore, among nine input factors, ‘pipe age’ and ‘location’ appeared insignificant whilst ‘pipe size’, ‘slope’, ‘the number of trees’ and ‘climatic condition’ were found consistently important over both models for serviceability deterioration process. The remaining three factors namely, ‘structure’, ‘soil’ and ‘buried depth’ might be redundant factors. A better and more consistent data collection regime may help to improve the predictive performance of the NNM and identify the significant factors.     

HCI ‘Intraface Model’ for System Design

We outline an alternative model of the interface in HCI, the ‘intraface’, in response to design issues arising from navigational and learning problems in hypertext domains. Ours is a model of general application to computer systems. It is composed of four key elements, identifiable within a dynamic interconnected context. These are the user; his/her interests; the tools employed and the ‘ensemble’ of representations brought to bear. In this paper we sketch the present shortcomings of HCI design before outlining the background for the model which draws upon two themes in contemporary psychology, conversational analysis and ‘affordance’ realist theories in perception. This framework allows for the development of principles of cooperation, user engagement and learning in HCI environments.     

Fuzzy logic techniques for mobile robot obstacle avoidance

This paper is concerned with the problem of reactive navigation for a mobile robot in an unknown clustered environment. We will define reactive navigation as a mapping between sensory data and commands. Building a reactive navigation system means providing such a mapping. It can come from a family of predefined functions (like potential fields methods) or it can be built using ‘universal’ approximators (like neural networks). In this paper, we will consider another ‘universal’ approximator: fuzzy logic. We will explain how to choose the rules using a behaviour decomposition approach. It is possible to build a controller working quite well but the classical problems are still there: oscillations and local minima. Finally, we will conclude that learning is necessary for a robust navigation system and fuzzy logic is an easy way to put some initial knowledge in the system to avoid learning from zero.     

Test feature selection for a complex electro-hydraulic servo system using frequency response measurements

A technique for the selection of the best set of test features for checkout or go-no-go test of a complex electro-hydraulic servo system from input-output measurements is presented. The first step is to establish checkout tolerance bands on the system response. Then the measurement set based on gain and phase which would best discriminate between the ‘healthy’ and ‘sick’ systems is selected from an initially large set of sampled frequencies via an optimization procedure in which the feature efficiency vector is introduced to add or discard features until a satisfactory set is obtained. Finally the selected feature sets are assessed for effectiveness via a ‘goodness’ criterion.     

The constituents of business interaction––generic layered patterns

In order to perform business modelling as apart of information systems development, there is a need for frameworks and methods. The paper proposes a framework for business interaction based on a language/action perspective. The framework is an architecture of five generic layers. The first layer concept is ‘business act’, which functions as the basic unit of analysis. The following four layer concepts are ‘action pair’, ‘exchange’, ‘business transaction’, and ‘transaction group’. The framework is inspired by a similar framework constructed by Weigand et al. The paper makes a critical examination of this framework as a basis for the proposed framework.     

Bioequivalence revisited: non-parametric analysis of two-period cross-over studies

Hauschke et al.'s non-parametric bioequivalence procedure for treatment effects and some aspects of computer implementation, among them Meineke and De Hey's algorithm, are explored. For studies with up to sixty subjects, a table of indices of the ranked intersubject-intergroup mean ratios or differences is given, to establish non-parametric 90% confidence intervals. It is shown that non-parametric analysis is not limited to treatment effects: it can also be applied to period and sequence effects. This extended procedure can be seen as the non-parametric analogue of analysis of variance on two-period cross-over studies. A FORTRAN program (BIOQNEW) incorporating Meineke and De Mey's algorithm is presented. This program provides non-parametric point estimates for treatment and period effects, 90% and 95% confidence intervals for test-versus-reference treatments, the 95% confidence interval for periods and a test on sequence effects, so that it can also be used for other than bioequivalence studies. BIOEQNEW can handle ratios (‘multiplicative model’) as well as differences (‘additive model’). It optionally provides the complete non-parametric posterior probability distribution for treatment ratios or differences, so that Schuirmann's ‘two one-sided tests procedure’ can also be performed in a non-parametric way.     

Simple association rules (SAR) and the SAR-based rule discovery

Association rule mining is one of the most important fields in data mining and knowledge discovery in databases. Rules explosion is a problem of concern, as conventional mining algorithms often produce too many rules for decision makers to digest. Instead, this paper concentrates on a smaller set of rules, namely, a set of simple association rules each with its consequent containing only a single attribute. Such a rule set can be used to derive all other association rules, meaning that the original rule set based on conventional algorithms can be ‘recovered’ from the simple rules without any information loss. The number of simple rules is much less than the number of all rules. Moreover, corresponding algorithms are developed such that certain forms of rules (e.g. ‘P?’ or ‘?Q’) can be generated in a more efficient manner based on simple rules.     

The least-disturbance principle and weak constraints

Certain problems, notably in computer vision, involve adjusting a set of real-valued labels to satisfy certain constraints. They can be formulated as optimisation problems, using the ‘least-disturbance’ principle: the minimal alteration is made to the labels that will achieve a consistent labelling. Under certain linear constraints, the solution can be achieved iteratively and in parallel, by hill-climbing. However, where ‘weak’ constraints are imposed on the labels — constraints that may be broken at a cost — the optimisation problem becomes non-convex; a continuous search for the solution is no longer satisfactory. A strategy is proposed for this case, by construction of convex envelopes and by the use of ‘graduated’ non-convexity.