No abstract

Traditionally, researchers have used either o-the-shelf models such as COCOMO, or developed local models using statistical techniques such as stepwise regression, to obtain software eort estimates. More recently, attention has turned to a variety of machine learning methods such as arti®cial neural networks (ANNs), case-based reasoning (CBR) and rule induction (RI). This paper outlines some comparative research into the use of these three machine learning methods to build software eort prediction systems. We brie¯y describe each method and then apply the techniques to a dataset of 81 software projects derived from a Canadian software house in the late 1980s. We compare the prediction systems in terms of three factors: accuracy, explanatory value and con®gurability. We show that ANN methods have superior accuracy and that RI methods are least accurate. However, this view is somewhat counteracted by problems with explanatory value and con®gurability. For example, we found that considerable eort was required to con®gure the ANN and that this compared very unfavourably with the other techniques, particularly CBR and least squares regression (LSR). We suggest that further work be carried out, both to further explore interaction between the enduser and the prediction system, and also to facilitate con®guration, particularly of ANNs. Ó

Patients with diabetes require annual screening for effective timing of sight-saving treatment. However, the lack of screening and the shortage of ophthalmologists limit the ocular health care available. This is stimulating research into automated analysis of the reflectance images of the ocular fundus. Publications applicable to the automated screening of diabetic retinopathy are summarised. The review has been structured to mimic some of the processes that an ophthalmologist performs when examining the retina. Thus image processing tasks, such as vessel and lesion location, are reviewed before any intelligent or automated systems. Most research has been undertaken in identification of the retinal vasculature and analysis of early pathological changes. Progress has been made in the identification of the retinal vasculature and the more common pathological features, such as small aneurysms and exudates. Ancillary research into image preprocessing has also been identified. In summary, the advent of digital data sets has made image analysis more accessible, although questions regarding the assessment of individual algorithms and whole systems are only just being addressed.

No abstract

This paper presents the design and evaluation of a novel, AI (Artificial Intelligence) based alarm processing and fault diagnosis system, for a 132kv/12 bus-16line sample power system. The work has been conducted in conjunction with Scottish Hydro Electric PLC. The fault diagnosis system is based on a hybrid object-oriented AI technique. The method developed utilises abductive inference.This technique is demonstrated to realise some improvements when compared with fuzzy logic and takes into account the current practical limitations in the design. The method is based on processing SCADA (Supervisory Control and Data Acquisition) messages, extending the arrangement of the knowledge acquisition process and applicability of circuit breakers and relays.The potential benefits and implications of adopting such an abductive fuuy knowledge based system are demonstrated in this research, and include a user friendly inference engine, adaptability, and KBS update.