1,721,014 research outputs found
Bayesian Conditioning for estimating the relative degrees of Reliability in a group of Neural Networks engaged at Iris Biometric Identification
No full textThe main problem with iris biometric identification systems is the presence of noises in the image of the eye (eyelid, eyelashes, etc...). To remove it many authors apply appropriate preprocessing to the image, but unfortunately this yields losses of information. Our work aims at correctly recognizing the subject also in presence of high rates of noise. The basic idea is that of partitioning the image of iris into 8 not-interleaved segments of the same size. Each segment is given to an LVQ network which generates prototypes with a high resistance to noise. Notwithstanding this, the 8 LVQ nets may still disagree in identifying the subject. In this paper we apply a method developed by the “belief revision” community to identify conflicts and rearrange the degrees of reliability of each expert (the LVQ nets) through a Bayesian algorithm. This estimated ranking of reliability is useful to take the final decision. Our work has produced an interesting 84 % of positive identificat..
Conflict Detection and Bayesian Conditioning for Estimating the Reliability of Each LVQ Network in a Group Engaged at Iris Biometric Identification
No full textThe main problem with iris biometric identification systems is the presence of noises in the image of the eye (eyelid, eyelashes, etc...). To remove it many authors apply appropriate preprocessing to the image, but unfortunately this yields losses of information. Our work aims at correctly recognizing the subject also in presence of high rates of noise. The basic idea is that of partitioning the image of iris into 8 not-interleaved segments of the same size. Each segment is given to an LVQ network which generates prototypes with a high resistance to noise. Notwithstanding this, the 8 LVQ nets may still disagree in identifying the subject. In this paper we apply a method developed by the "belief revision" community to identify conflicts and rearrange the degrees of reliability of each expert (the LVQ nets) through a Bayesian algorithm. This estimated ranking of reliability is useful to take the final decision
Non-isometric colour similarity
No full textThis paper presents the findings obtained in an experimental study of metric underlying the perceptual colour space. Previous studies evidenced that, in tasks of evaluation of colour similarity, the subjects don't refer to the most general category of “colour”, but rather rely on the introduction of subordinate categories containing all variations of a colour. Besides, categories related to different colours can sometimes overlap. This forces to conclude that perception of colour variations is not isometric, but is rather weighed in different ways for different colours.
In order to detect the metric of colour space we performed an experiment with multiple conditions within the subjects, designed to determine the form of the function that ties the independent variable (tonality of colour) with the dependent variable (similarity judgement). To each subject we presented simultaneously a pair of images, the target one and another differing from the target only for its colour (a suitable perturbation of the tonality). The subject task was to rate the similarity of the second image with the target. The frequency distribution of similarity judgments for each colour gave a qualitative description of how the different colours are represented at the cognitive level. We applied to the observed frequencies a unidimensional scaling procedure to obtain a precise measure of the distance between the variation steps for each colour. We were allowed to choose a single dimension because we limited the study only to the variation of tonality. The scaling was applied separately to each colour scale.
The results showed that different colours were associated to different measure scales. Besides, once chosen a particular colour, its measure scale itself was depending on the direction of variation chosen for its tonality during the experimental presentation. We can conclude that the geometry of colour space looks very complicated and not reducible to familiar mathematical concepts
Feedback control tames disorder in attractor neural networks
No full textTypical attractor neural networks (ANN) used to model associative memories behave like disordered systems, as the asymptotic state of their dynamics depends in a crucial (and often unpredictable) way on the chosen initial state. In this paper we suggest that this circumstance occurs only when we deal with such ANN as isolated systems. If we introduce a suitable control, coming from the interaction with a reactive external environment, then the disordered nature of ANN dynamics can be reduced, or even disappear. To support this claim we resort to a simple example based on a version of Hopfield autoassociative memory model interacting with an external environment which modifies the network weights as a function of the equilibrium state coming from retrieval dynamics
Systemic Analysis of Normal and Pathological EEG Signal: Clinical Applications and Methodological Limits
No full text
THE EMERGENCE OF PERCEPTUAL GROUPING ABILITY IN NEURAL NETWORK MODELS OF VISUAL CORTEX OPERATION
No full text- …
