RESUMEN

In glass bottle inspection, the defects detection is of first importance. For online system detection, high speed and robust detection of faults are highly required. Neural networks have recently, and successfully, been applied to fault detection in many manufacturing processes. In this study, a Gaussian neural network, an extension of the RBF network, trained through a competitive algorithm, has been chosen for fault detection. Four parameters extracted from images of the bottles are used as inputs of the network. The number of Gaussian units is adjusted by an informational criterion. Experimental results show that the performance of this network are better than classical parametric and non parametric classifiers.

Asunto(s)

Asunto(s)

RESUMEN

A major problem in mass screening for glaucoma is the substantial time required to detect areas of depressed visual sensitivity, scotoma. We have developed a novel instrument for testing rapidly and automatically the sensitivity of the visual field. The patient views a large CRT screen on which up to four luminous points are presented in a prearranged pattern. The patient's vocal response is a number from zero to four, the number of stimulus points that he has observed, which is recognized by the controller and recorded. A series of 90 such patterns is presented sequentially, giving a 288 points visual field test, at the conclusion of which a hard copy probabilistic map of the retinal location of scotoma is drawn. The examination procedure is much faster than that using conventional 'automatic' tests and its results are almost identical with those from two commercial instruments with which our experimental 'perimeter' has been compared. Initial clinical trials are very encouraging.

Asunto(s)

RESUMEN

A fast algorithm for the well-known Parzen window method to estimate density functions from the samples is described. The computational efforts required by the conventional and straightforward implementation of this estimation procedure limit its practical application to data of low dimensionality. The proposed algorithm makes the computation of the same density estimates with a substantial reduction of computer time possible, especially for data of high dimensionality. Some simulation experiments are presented which demonstrate the efficiency of the method. They indicate the computational savings that may be achieved through the use of this fast algorithm for artificially generated sets of data.

RESUMEN

In this paper, an approach to unsupervised pattern classifiation is discussed. The classification scheme is based on an approximation of the probability densities of each class under the assumption that the input patterns are of a normal mixture. The proposed technique for identifying the mixture does not require prior information. The description of the mixture in terms of convexity allows to determine, from a totally unlabeled set of samples, the number of components and, for each of them, approximate values of the mean vector, the covariance matrix, and the a priori probability. Discriminant functions can then be constructed. Computer simulations show that the procedure yields decision rules whose performances remain close to the optimum Bayes minimum error-rate, while involving only a small amount of computation.

Asunto(s)

RESUMEN

To investigate the nature of variations in the large intestine potential differences, a continuous perfusion of isotonic saline was carried out in the colon of 14 rats. Intraluminal pressure and potential differences between the lumen and the peritoneal cavity were continuously and simultaneously recorded, while impedance of the system and respiration were also constantly monitored. To obtain a quantitative evaluation of the data, Fast Fouier Transform was performed on the signals and their derivatives which were auto- and cross-correlated. While there was no obvious relation between pressure and potential in the unperfused colon, there was clear visual qualirative evidence that, during steady state conditions of perfusion, an increase in intraluminal pressure was accompanied by a decrease in potential differences, while impedance of the recording system remained unchanged. Computer analysis disclosed four narrow ranges of stable frequencies for both pressure and potential. They were centred around 0-3, 1-75, 10-7, and 75 cycles per minute, the latter being synchronous with respiration. It is concluded that the variations of potential differences recorded during perfusion, a well-know phenomenon, are not electrical artefacts: the fast rhythm is probably induced by respiration, which increases intracolonic pressure and that, in turn, reduces the absolute value of potential differences, which remain negative mucosa versus serosa. The slower rhythms are synchronous for pressure and potential. Mechanisms responsible for the decrease in potential related to the increase in pressure remain unknown.

Asunto(s)

Asunto(s)