Alexander V. Goltsev

An assembly neural network for texture segmentation

Abstract An architecture of a neural network with assembly organization is described. Such network architecture is applied to the problem of texture segmentation in natural scenes. The network is partitioned into several subnetworks. Each subnetwork is a column structure in which features are represented by means of “float” coding. Input data excite corresponding “floats” of neurons in the subnetworks. In the process of learning the weights of modifiable connections between excited neurons are changed so that Hebbs assemblies are formed in the column structures. All subnetworks are incorporated into a single network by a neural activity control system. Computer simulation of the proposed network has been performed. The results of computer simulations show the possibility of successful application of the assembly neural network to the problem of texture segmentation.

Investigation of efficient features for image recognition by neural networks

In the paper, effective and simple features for image recognition (named LiRA-features) are investigated in the task of handwritten digit recognition. Two neural network classifiers are considered-a modified 3-layer perceptron LiRA and a modular assembly neural network. A method of feature selection is proposed that analyses connection weights formed in the preliminary learning process of a neural network classifier. In the experiments using the MNIST database of handwritten digits, the feature selection procedure allows reduction of feature number (from 60 000 to 7000) preserving comparable recognition capability while accelerating computations. Experimental comparison between the LiRA perceptron and the modular assembly neural network is accomplished, which shows that recognition capability of the modular assembly neural network is somewhat better.

Combination of the assembly neural network with a perceptron for recognition of handwritten digits arranged in numeral strings

The purpose of the paper is to design and test neural network structures and mechanisms for making use of the information that is contained in the character strings for more correct recognition of the characters constituting these strings. Two neural networks are considered in the paper; both networks are combined into a joint recognition system. The first is the assembly neural network and the second is the neural network of a perceptron type. A computer simulation of the system is performed. The combined system solves the task of recognition of handwritten digits of the MNIST test set provided that the digits have been arranged in the numeral strings memorized in the system. During a recognition process of an input numeral string, the assembly neural network executes intermediate recognition of the digits basing on which a perceptron type network accomplishes the final choice among the limited combinations of strings memorized in the network. The experiments have demonstrated that the combined system is able to make use of the information that is contained in the strings for more correct recognition of digits of the MNIST test set. In particular, the experiments have shown that the combined system commits no errors in the recognition of MNIST test set on the condition that the digits of this set had been organized in the strings of more than 5 digits each.

Inhibitory connections in the assembly neural network for texture segmentation

A neural network with assembly organization is described. This assembly network is applied to the problem of texture segmentation in natural scenes. The network is partitioned into several subnetworks: one for each texture class. Hebbs assemblies are formed in the subnetworks during the process of training the excitatory connections. Also, a structure of the inhibitory connections is formed in the assembly network during a separate training process. The inhibitory connections result in inhibitory interactions between different subnetworks. Computer simulation of the network has been performed. Experiments show that an adequately trained assembly network with inhibitory connections is more efficient than without them.

GENERALIZATION OF FEATURES IN THE ASSEMBLY NEURAL NETWORKS

The purpose of the paper is an experimental study of the formation of class descriptions, taking place during learning, in assembly neural networks. The assembly neural network is artificially partitioned into several sub-networks according to the number of classes that the network has to recognize. The features extracted from input data are represented in neural column structures of the sub-networks. Hebbian neural assemblies are formed in the column structure of the sub-networks by weight adaptation. A specific class description is formed in each sub-network of the assembly neural network due to intersections between the neural assemblies. The process of formation of class descriptions in the sub-networks is interpreted as feature generalization. A set of special experiments is performed to study this process, on a task of character recognition using the MNIST database.

Modular neural networks with Hebbian learning rule

The paper consists of two parts, each of them describing a learning neural network with the same modular architecture and with a similar set of functioning algorithms. Both networks are artificially partitioned into several equal modules according to the number of classes that the network has to recognize. Hebbian learning rule is used for network training. In the first network, learning process is concentrated inside the modules so that a system of intersecting neural assemblies is formed in each module. Unlike that, in the second network, learning connections link only neurons of different modules. Computer simulation of the networks is performed. Testing of the networks is executed on the MNIST database. Both networks directly use brightness values of image pixels as features. The second network has a better performance than the first one and demonstrates the recognition rate of 98.15%.

Brightness image segmentation by a neuron-like network

Abstract An algorithm for brightness image segmentation is described. The algorithm is realized as computer model of a layered neural network. The result of the models run on real images is presented.

A Process of Differentiation in the Assembly Neural Network

An assembly neural network model is described. The network is artificially partitioned into several sub-networks according to the number of classes that the network has to recognize. In the process of primary learning Hebb’s neural assemblies are formed in the sub-networks by means of modification of connections’ weights. Then, a differentiation process is executed which significantly improves the recognition accuracy of the network. A computer simulation of the assembly network is performed with the aid of which the differentiation process is studied in a set of experiments on a character recognition task using two types of separate handwritten characters: Ukrainian letters and Arabic numerals of MNIST database.

A RECURRENT NEURAL NETWORK FOR PARTITIONING OF HAND DRAWN CHARACTERS INTO STROKES OF DIFFERENT ORIENTATIONS

A neural network is described which is intended to extract orientation features that should be used for recognition of hand drawn characters. The network partitions the input hand drawn characters into separate line segments (strokes) according to their orientations. The network consists of several neural layers; each layer serves for extracting strokes of a certain orientation. Every neural layer has one-to-one correspondence with an input screen. The network uses an iterative update procedure which includes interactions of neurons inside each layer through oriented excitatory connections and inhibitory interrelations between the corresponding neurons of different layers. Computer simulation of the network was performed. Experiments showed that the network efficiently classifies all pixels of any hand drawn characters according to the orientations of the strokes constituting these characters and performs, as a result of that, a reasonable segmentation of characters.

Extraction of homogeneous fine-grained texture segments in visual images

A new heuristic algorithm is proposed for extraction of all homogeneous fine-grained texture segments present in any visual image. The segments extracted by this algorithm should comply with human understanding of homogeneous fine-grained areas. The algorithm sequentially extracts segments from more homogeneous to less homogeneous ones. The algorithm belongs to a region growing approach. So, for each segment, an initial seed point of this segment is found. Then, from this initial pixel, the segment begins to expand occupying its adjacent neighborhoods. This procedure of expansion of the segment continues till the segment reaches its borders. The algorithm examines neighboring pixels using texture features extracted in the image by means of a set of texture windows. The segmentation process terminates when the image contains no more sizable homogeneous segments. The segmentation procedure is fully unsupervised, i.e., it does not use a priori knowledge on either the type of textures or the number of texture segments in the image. Using black and white natural scenes, a series of experiments demonstrates efficiency of the algorithm in extraction of homogeneous fine-grained texture segments and the segmentation looks reasonable “from a human point of view”.