Use of nonmonotonic inheritance networks for the storage and maintenance of knowledge discovered in data is presented. When nonmonotonic inheritance networks are used instead of rules extracted from identification tre...
详细信息
Use of nonmonotonic inheritance networks for the storage and maintenance of knowledge discovered in data is presented. When nonmonotonic inheritance networks are used instead of rules extracted from identification trees for storing knowledge discovered in data, additional conflicting information can be inserted directly into such networks, thereby by-passing the need for recompilation. The generation of a nonmonotonic inheritance network provides an intermediate representation between the original identification tree and an expression/rule-based representation of the knowledge extracted from the data. The path-based preemption algorithms presented are probably correct and have the advantage of being able to cope with conflicting conclusions.
To design, implement and operate an efficient manufacturing organizations today, a systems view is necessary;this picture is further complicated as manufacturing systems are also dynamic in nature. Holistic design of ...
详细信息
To design, implement and operate an efficient manufacturing organizations today, a systems view is necessary;this picture is further complicated as manufacturing systems are also dynamic in nature. Holistic design of manufacturing systems, that is using dynamic systems thinking based on business strategy, can help to optimize technical, organizational and human factors in new system implementation. During the last decade numerous packages for the simulation of manufacturing systems have been developed as a support for the analysis and design of these systems. Simulation has traditionally concentrated on details, which is important at the design stage, but many of the total system aspects such as machine, organization and human aspects are lost. This paper presents the concept of a virtual arena as a means of approaching the holistic design of manufacturing systems;the literature relating to this concept is also presented in outline. Special interest is placed upon work relating to the total or holistic view of manufacturing organizations. Current research initiatives in Scandinavia are also presented.
The British Army has a long-term goal of providing a federation of interoperable systems so as to give the users the impression of a single and fully integrated Army CIS for use in peace and war, on an off the battlef...
详细信息
The British Army has a long-term goal of providing a federation of interoperable systems so as to give the users the impression of a single and fully integrated Army CIS for use in peace and war, on an off the battlefield. Effective data management is seen as essential if interoperability is to be achieved. This paper identifies that the way to achieve effective data management is to build a single Logical Data Model for the whole of the Army, with the model being built by projects collaborating with the data management organization to expand a central core. This central core is based on the requirements of the staff of an operational headquarters. The major problem to be overcome in a collaborative exercise of this kind is the data modeller's natural inclination to demonstrate his artistry by producing his model without benefit of previous work. This implies the necessity of effecting a change in the prevailing culture within the data modelling community.
New and proposed legislation is having a major effect on the management of health and safety within the construction industry. When carrying out and planning activities on-site, considerable expertise is required to e...
详细信息
New and proposed legislation is having a major effect on the management of health and safety within the construction industry. When carrying out and planning activities on-site, considerable expertise is required to ensure that the work is done safely, and a thorough knowledge and understanding of the relevant legislative documentation is necessary. A prototype computer system has been built to assist those responsible for safety management. A framework has been constructed which allows this expertise to be encoded as rules within a knowledge-based system. Where the actions that may be taken are constrained by legislation, rules are placed within the knowledge-base to reflect these constraints.
Increasingly, modern electronic systems are required to be mixed hardware/software systems. Traditional system development techniques do not cope with the needs of complex systems, and in particular with rapidly chang...
详细信息
Increasingly, modern electronic systems are required to be mixed hardware/software systems. Traditional system development techniques do not cope with the needs of complex systems, and in particular with rapidly changing requirements under time-to-market pressure. Hardware/Software codesign techniques offer considerable help in that they allow one to view an application as a flexible composition of properly encapsulated hardware and software components. Some of these components could be existing adaptable and reusable library components. This paper addresses the problem of hardware/software codesign when starting at the system-level. The input is a functional specification given in an executable language such as SDL. The codesign involves a number of interactive steps leading to an implementation composed of hardware and software components. The design decisions and refinement steps are made using an intermediate form called Solar. As shown in Figure 1, the steps needed to go from a functional specification to an implementation include system-level partitioning, communication synthesis, and virtual prototyping.
Evidence theory shows promise in uncertainty handling for knowledge discovery in databases. It allows an expression for ignorance which is an intuitive way for dealing with missing values. The ability to combine diffe...
详细信息
Evidence theory shows promise in uncertainty handling for knowledge discovery in databases. It allows an expression for ignorance which is an intuitive way for dealing with missing values. The ability to combine different pieces of evidence at various levels of coarseness has clear advantages in discovery of knowledge from distributed, parallel and heterogeneous databases. It allows the development of parallel algorithms for database mining. The framework for knowledge discovery in databases based on evidence theory also allows the incorporation of domain knowledge into the discovery process by using evidential operators. The framework is parallel and discovery processes using this framework is parallel, making them efficient for large data sets.
Embedded system design faces the HW/SW-partitioning problem. This is often solved by explicit designer decisions. In this paper we describe a partitioning method based on design specification analysis. Static and dyna...
详细信息
Embedded system design faces the HW/SW-partitioning problem. This is often solved by explicit designer decisions. In this paper we describe a partitioning method based on design specification analysis. Static and dynamic aspects influence the overall time behavior and are examined during two analysis phases. Each phase evaluates a cost function and classifies the suitability of the regarded partition for HW-implementation. Furthermore costs of the HW/SW-interface are taken into account because this overhead may not be neglected. The analysis method is demonstrated by several benchmarks and reasonable results are pointed out. It has also been pointed out that the efficiency of embedded systems is dramatically influenced by the interface between HW- and SW. We developed a pipelined interface using the features of the target architecture. The interface allows handling of multiple HW-components and parameter transfer is quick and flexible.
The Database Mining Research Group within the School of Information and Software Engineering, at the University of Ulster, Jordanstown, has set two main objectives for its work. The first objective is to devise improv...
详细信息
The Database Mining Research Group within the School of Information and Software Engineering, at the University of Ulster, Jordanstown, has set two main objectives for its work. The first objective is to devise improved data mining algorithms for mining classification and association rules. Some of the areas of improvement being investigated are: more efficient algorithms;wider algorithm scope;noisy data handling;incorporation of prior knowledge into algorithms;and measures of interestingness. The second objective is to evaluate the algorithms on real world data sets. The use of data mining techniques on data sets from the Northern Ireland Housing Executive (NIHE) is also investigated.
Databases are complex structures that may conceal implicit patterns of information that cannot be easily discovered by conventional analysis and interrogation methods. This situation can be exacerbated as the database...
详细信息
Databases are complex structures that may conceal implicit patterns of information that cannot be easily discovered by conventional analysis and interrogation methods. This situation can be exacerbated as the database grows in size, and the data therein grows in complexity. Discovery of patterns and trends in such cases requires database query methods more advanced than those traditionally used. Such databases may be analyzed using a set of techniques often collectively referred to as knowledge discovery. Knowledge discovery may be performed effectively using neural networks. The use of neural network techniques for knowledge discovery in an infertility database is described.
The principal constituents of soft computing are the fuzzy logic (FL), artificial neural networks (NN) and probabilistic reasoning (PR). It is generally regarded that FL primarily deals with imprecision, NN with learn...
详细信息
The principal constituents of soft computing are the fuzzy logic (FL), artificial neural networks (NN) and probabilistic reasoning (PR). It is generally regarded that FL primarily deals with imprecision, NN with learning, and PR with uncertainty. They have, however, overlapping boundaries and are known to be complementary rather than competitive to each other in many applications. Here, two control algorithms, one implemented by fuzzy logic and the other by a neural network, are used as the basis to highlight salient features of soft computing. A DC motor servo system with the proposed soft computing based algorithms is discussed. The fuzzy logic control employs the principles of fuzzy logic to calculate an optimal output action based on input conditions, and a knowledge base expressed in linguistic forms, thereby performing a parallel operation to control the output with a high degree of robustness against parameter change. In the neural network control, focus is on how neural networks can overcome deadzone-plus-saturation nonlinearity commonly found in the power driver of a DC servo motor. Simulation results have been performed to establish the validity of these control algorithms.< >
暂无评论