An inverse problem in neural processing.

M N Oğuztöreli; T M Caelli

doi:10.1007/bf00336994

Home/Archive/PubMed

PubMed · 3955099

An inverse problem in neural processing.

M N Oğuztöreli·T M Caelli

View original source ↗Export metadata ↓

Abstract

Any neural network aimed at the coding sensory events must contain computational properties which generally allow the organism to reconstruct the input signals with some degree of accuracy--else the association between stimulus and response would, at best, be uncertain. In this paper we investigate the problem of reconstructing external input signals to neural networks when the activity profiles of only some of its member cells are known. The evolution and activities of such cells are defined by an earlier formulation of one of us (Oğuztöreli 1979) and, here, we restrict our application to local circuits within the vertebrate retina. Solutions to this inverse coding problem are presented for specific network equations and exemplified with 1, 3, and 5 neuron cases.

Explore related subjects

Mathematics Models, Neurological Neural Pathways Neurons, Afferent Retina Journal Article Research Support, Non-U.S. Gov't

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

M N Oğuztöreli, T M Caelli. 1986. An inverse problem in neural processing.. https://doi.org/10.1007/bf00336994

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related citations

PubMed1999

Improved multidimensional scaling analysis using neural networks with distance-error backpropagation.

We show that neural networks, with a suitable error function for backpropagation, can be successfully used for metric multidimensional scaling (MDS) (i.e., dimensional reduction while trying to preserve the original distances between patterns) and are in fact able to outdo the standard algebraic approach to MDS, known as classical scaling.

Mathematics↗

PubMed1998

Geometry of circular vectors and pattern recognition of shape of a boundary.

This paper deals with pattern recognition of the shape of the boundary of closed figures on the basis of a circular sequence of measurements taken on the boundary at equal intervals of a suitably chosen argument with an arbitrary starting point. A distance measure between two boundaries is defined in such a way that it has zero value when the associated sequences of measurements coincide by shifting the starting point of one of the sequences. Such a distance measure, which is invariant to the starting point of the sequence of measurements, is used in identification or discrimination by the shape of the boundary of a closed figure. The mean shape of a given set of closed figures is defined, and tests of significance of differences in mean shape between populations are proposed.

Mathematics↗

PubMed1998

A simple method to verify in vivo the accuracy of target coordinates in linear accelerator radiosurgery.

PURPOSE: A simple method that verifies the coincidence of the isocenter with the center of the target volume in radiosurgery treatment conditions is described. The accuracy is compared to that of accepted computerized procedures employing fiducial markers. METHODS AND MATERIALS: The center of the beam is identified by a cylindrical localizer, fixed to the plate of the supplemental collimator, with a 2 x 50 mm tungsten rod coincident with the beam axis and is projected onto the x-ray portal verification films. Prior to irradiation, the coordinates of the intersection of the beams axes, which is in a known spatial relationship with the isocenter, are read directly on portal x-ray films and their coincidence with the coordinates set during patient positioning, is checked. RESULTS: The mean displacement in AP, Lat, and Vert coordinates respectively, over 84 patients, between the coordinates calculated by the computerized procedure employing fiducial markers and the coordinates calculated by using the rulers was 0.3 +/- 0.4 mm. CONCLUSIONS: From the results obtained with the two methods we can conclude that rulers method can be used as a fast indirect control of the position of the radiation isocenter. Moreover, the dimensions of the radiation field and the correct alignment of the tertiary circular collimator can be also documented.

Mathematics↗