Search

Outliers are difficult to handle because some of them can be measurement errors, while others may represent phenomena of interest, something "significant" from the viewpoint of the application domain. Statistical and computational methods have been proposed to detect outliers, but further analysis of outliers requires much relevant domain knowledge. In our previous work (1994), we suggested a knowledge-based method for distinguishing between the measurement errors and phenomena of interest by modeling "real measurements" - how measurements should be distributed in an application domain. In this paper, we make this distinction by modeling measurement errors instead. This is a cautious approach to outlier analysis, which has been successfully applied to a medical problem and may find interesting applications in other domains such as science, engineering, finance, and economics.

A software-based visual-field testing (perimetry) system is described which incorporates several AI components, including machine learning, an intelligent user interface and pattern discovery. This system has been successfully used for self-screening in several different public environments.

The pitting behavior of passive 304 stainless steel in 0.25 M Na 2 SO 4 media of pH 1 containing chloride ions of different concentrations has been investigated using EIS technique. The characteristics of measured impedance spectra have been analysed. It is proposed that an intermediate complex (MOHCl)com will form in both the pit initiation and propagation stages. The reaction mechanism models for pit initiation and propagation, and an equivalent circuit model to simulate the electrode impedance in pit propagation stage have been suggested. The factors which control the pit growth kinetics have been discussed.

One of the main difficulties in obtaining reliable data from patients in glaucomatous tests is the measurement noise caused by the learning effect, inattention, failure of fixation, fatigue, etc. Using Kohonen's self-organising feature maps, we have developed a computational method to distinguish between the noise and true measurement. This method has been shown to provide a satisfactory way of locating and rejecting noise in the test data, an improvement over conventional statistical methods. >

The effects of the substitutions of Dy and Er for Sm on the crystal structure and magnetic properties of Sm/sub 2/Fe/sub 17/ and Sm/sub 2/Fe/sub 17/N/sub y/ compounds have been investigated. The structures of (Sm/sub 1-x/R/sub x/)/sub 2/ Fe/sub 17/N/sub y/ compounds with R=Dy and Er and the hosts change from Th/sub 2/Zn/sub 17/-type to Th/sub 2/Ni/sub 17/-type with increasing Dy and Er contents. The replacement of Sm by Dy or Er results in a decrease in Curie temperature, saturation magnetization, and magnetic anisotropy field in the (Sm/sub 1-x/R/sub x/)/sub 2/Fe/sub 17/N/sub y/ compounds. A tentative spin phase diagram of the (Sm/sub 1-x/Dy/sub x/)/sub 2/Fe/sub 17/N/sub y/ compounds is suggested. >

Diagnosis of visual function losses in glaucomatous patients depends to a large extent on the analysis of the data collected from corresponding psychophysical tests. One of the main difficulties in obtaining reliable data from patients in these tests is the measurement noise caused by the learning effect, inattention, failure of fixation, fatigue etc. Using Kohonen's self-organising feature map, we have developed a computational method to distinguish between the noise and true measurement and to provide an instant assessment of reliability of the computer-based visual function test. In particular we have experimented with 270 test records from glaucoma patients and glaucoma suspects and found that this method provides a satisfactory way of locating and rejecting noise in the test data, an improvement over conventional statistical methods. This method can also provide doctors with a clear view of the patient's behaviour during the test, thus assisting in their diagnostic decision making process.