Tatiana Baidyk

Rosenblatt perceptrons for handwritten digit recognition

The Rosenblatt perceptron was used for handwritten digit recognition. For testing its performance the MNIST database was used. 60,000 samples of handwritten digits were used for perceptron training, and 10,000 samples for testing. A recognition rate of 99.2% was obtained. The critical parameter of Rosenblatt perceptrons is the number of neurons N in the associative neuron layer. We changed the parameter N from 1,000 to 512,000. We investigated the influence of this parameter on the performance of the Rosenblatt perceptron. Increasing N from 1,000 to 512,000 involves decreasing of test errors from 5 to 8 times. It was shown that a large scale Rosenblatt perceptron is comparable with the best classifiers checked on MNIST database (98.9%-99.3%).

Neural network system for face recognition

An image recognition method based on neural network system is proposed. This method uses the permutative coding technique for image preprocessing and neural classifier for image recognition. We have proposed the permutative coding technique to make recognition process invariant to small displacements of the object in the image. The system was tested on the ORL database. This database contains 400 face images of 40 persons. 200 images are used for training and 200 for recognition. The error rate of 0.1% for face recognition was obtained. This method was tested also with 40, 80, 120 and 160 images for system training and the rest images for recognition. The error rates 16.1%, 7.09%, 2.15% and 1.4% were obtained respectively.

Neural Networks and Micromechanics

This is an interdisciplinary field of research involving the use of neural network techniques for image recognition applied to tasks in the area of micromechanics. The book is organized into chapters on classic neural networks and novel neural classifiers; recognition of textures and object forms; micromechanics; and adaptive algorithms with neural and image recognition applications. The authors include theoretical analysis of the proposed approach, they describe their machine tool prototypes in detail, and they present results from experiments involving microassembly, and handwriting and face recognition. This book will benefit scientists, researchers and students working in artificial intelligence, particularly in the fields of image recognition and neural networks, and practitioners in the area of microengineering.

Image Recognition Systems Based on Random Local Descriptors

Two image recognition systems based on random local descriptors are described. Random local descriptors play the role of features that have to be extracted from the image. The advantage of this type of features is a possibility to create sufficiently general description of the image. This approach was tested in different image recognition tasks: handwritten digit recognition, face recognition, metal surface texture recognition and micro work piece shape recognition. The best result for handwritten digit recognition on the MNIST database is the error rate of 0.37% and for face recognition on the ORL database is the error rate of 0.1%. The results for texture and micro work piece shape recognition are also promising.

Limited receptive area neural classifier for texture recognition of mechanically treated metal surfaces

The limited receptive area (LIRA) neural classifier is proposed for texture recognition of mechanically treated metal surfaces. It may be applied in systems that have to recognize position and orientation of complex work pieces during micromechanical device assembly as well as in surface quality inspection systems. The performance of the proposed classifier was tested on a specially created image database with four texture types corresponding to metal surfaces after milling, polishing with sandpaper, turning with lathe and polishing with file. The promising recognition rate of 99.8% was obtained.

Recognition of Swallowing Sounds Using Time-Frequency Decomposition and Limited Receptive Area Neural Classifier

In this paper we propose a novel swallowing sound recognition technique based on the limited receptive area (LIRA) neural classifier and time-frequency decomposition. Time-frequency decomposition methods commonly used in sound recognition increase dimensionality of the signal and require steps of feature selection and extraction. Quite often feature selection is based on a set of empirically chosen statistics, making the pattern recognition dependent on the intuition and skills of the investigator. A limited set of extracted features is then presented to a classifier. The proposed method avoids the steps of feature selection and extraction by delegating them to a limited receptive area neural (LIRA) classifier. LIRA neural classifier utilizes the increase in dimensionality of the signal to create a large number of random features in the time-frequency domain that assure a good description of the signal without prior assumptions of the signal properties. Features that do not provide useful information for separation of classes do not obtain significant weights during classifier training. The proposed methodology was tested on the task of recognition of swallowing sounds with two different algorithms of time-frequency decomposition, short-time Fourier transform (STFT) and continuous wavelet transform (CWT). The experimental results suggest high efficiency and reliability of the proposed approach.

Permutative coding technique for handwritten digit recognition system

The new neural classifier for the handwritten digit recognition is proposed. The classifier is based on the Permutative Coding technique. This coding technique is derived from the associative-projective neural networks developed in the 80th-90th. The classifier performance was tested on the MNIST database. The error rate of 0.54% was obtained.

Micro-facet solar concentrator

A low-cost micro-facet solar concentrator is proposed. A large number of small flat mirrors are situated in a parabolic surface to approximate a large parabolic mirror. Low-cost commercial flat mirrors can be used for manufacturing such concentrators. Geometrical analyses show that this concentrator will have a concentration rate of some hundreds of suns. The problems of production of micro mirrors, support components, and the automatic assembly of the concentrator are discussed. Rough estimations show that the cost of the concentrator should be ∼

LIRA neural classifier for handwritten digit recognition and visual controlled microassembly

55 per square metre of concentrator surface.

Pairwise Permutation Coding Neural Classifier

Abstract In this paper, limited receptive area neural classifiers are described which are based upon Rosenblatts perceptron. These networks can be used for both binary and gray-level images. A method is reviewed for greatly expanding the amount of available training data. A training algorithm, based upon that of Rosenblatt, is given. The networks are applied to the handwritten numeral recognition problem, and to task in microassembly image recognition.