No abstract

Вінницький національний технічний університет, м. Вінниця Анотація. Використання нейромережевих технологій при створенні експертних систем показало їх перспективність для медичного діагностування захворювань при значній кількості симптомів. Базова модель нейромережевого класифікатора зумовила використання дискримінантного аналізу для процесу класифікації. В роботі розглянуто вдосконалений варіант класифікатора на базі нейромережі Хеммінга. Вдосконалення стосується усунення одного зі зворотних латеральних зв'язків у кожного нейроподібного елемента останнього шару нейромережі. Це призвело до спрощення структури класифікатора. Імітаційне моделювання класифікаційного процесу проводилось на прикладах з медичного діагностування. Воно показало проскорення процесу класифікації майже у 2 рази у порівнянні з класичним варіантом цього процесу. Ключові слова: нейромережевий класифікатор, комп'ютерне моделювання, латеральний зв'язок, медичне діагностування. Аннотация. Использование нейросетевых технологий при создании экспертных систем показало их перспективность для медицинского диагностирования заболеваний при значительном количестве симптомов. Базовая модель нейросетевого классификатора обусловила приминение дискриминантного анализа для процесса классификации. В работе рассмотрен усовершенствованный вариант классификатора на базе нейросети Хэмминга. Усовершенствование касается устранения одного из обратных латеральных связей у каждого нейроподобного элемента последнего слоя нейросети. Это привело к упрощению структуры классификатора. Имитационное моделирование классификационного процесса проводилось на примерах из медицинского диагностирования. Оно показало ускорение процесса классификации почти в 2 раза по сравнению с классическим вариантом этого процесса. Ключевые слова: нейросетевой классификатор, компьютерное моделирование, латеральная связь, медицинское диагностирование. Abstract. The use of neural network technologies in the creation of expert systems has shown their promise for medical diagnosis of diseases with a significant number of symptoms. The basic model of the neural network classifier has led to the use of discriminant analysis for the classification process. The paper considers the advanced version of the classifier based on Hamming's neural network. Improvement relates to the elimination of one of the reverse lateral bonds in each neural-like element of the last layer of the neural network. This led to a simplification of the classifier's structure. Simulation modeling of the classification process was carried out on examples of medical diagnosis. It showed an acceleration of the classification process by almost 2 times compared with the classic version of this process.

No abstract

Social media is becoming increasingly used as a source of information, including events during warfare. The fake accounts of the social media are often used for a variety of cyber-attacks, information-psychological operations, and social opinion manipulating during warfare. The analysis of online social media research methods are carried out, the main metrics and attributes of fake accounts in Facebook are investigated. Each metric is assigned to the appropriate categories for the convenience of their analysis and gets a certain number of points depending on conditions from 0 to 3, which indicate how much every of the metrics influenced on conclusion about the fakeness of the account. The levels of influence have the following meanings: 0 – no influence, 1 – weak influence, 2 – significant influence, 3 – critical influence. For example, if the histogram feature reaches level 3, this means that the parameter characterizing this feature has a critical impact on account fakeness. Otherwise, if the column is at 0 or 1 level, this means that the parameter is inherent in the real account. Thus, based on the level of each of the parameters, we conclude on the fakeness or reality of a certain account. The following metrics are analyzed: likes, friends, posts and statuses, personal information about the user and the photos, considering their possible parameters and influence on the status of the account. Each metric is assigned to the appropriate categories for the convenience of their analysis. A decision-making system based on a supported vector machine is developed and has 9 inputs and single output. A series of experimental research was conducted where account analyzing as well as parameters extracting and selection are realized on Facebook. The classifier accuracy of the fake accounts detection is 97% with the special prepared dataset of the real and fake account parameters.

The subject of study in this article is the features of structural organization and functioning of the improved Hamming network as a model of neural network heteroassociative memory for classification by discriminant functions. The goal is to improve the neural network classifier based on the Hamming network, which implements the criterion of maximum similarity using discriminant functions and does not have restrictions on the representation of input data (not only binary data). The tasks: analyze the capabilities of associative memory models using neural networks as an example; analyze the features of classification on the principles of discriminant analysis; develop the structure of a neural network classifier as a model of neural network heteroassociative memory; perform simulation modeling of the classification process on the example of medical diagnosis. The methods used are a mathematical model of the functioning of a neural network as a classifier, and simulation in C#. The following results have been obtained: the structure of the neural network classifier has been improved through the formation connection matrix of a hidden layer from pre-calculated coefficients of linear discriminant functions, and the connection matrix of the output layer in the form symmetrical matrix with zeros on the main diagonal. This allows not only to simplify m connections, where m is the number of classes, in the structure of the output layer of the neural network classifier, but also to speed up the classification process, as well as to implement classification by the maximum of discriminant functions. Conclusions. The scientific novelty of the results obtained is as follows: the neural network classification method has been improved using pre-calculated elements of the connection matrices in the hidden and output layers of the classifier, which does not imply a long process of direct neural network learning with using discriminant functions; the structural organization of a neural network classifier is proposed, which is an improvement of the Hamming network as a model of heteroassociative memory, that allows using this classifier in a decision support system for medical diagnosis; the removal of positive feedback in neurons of the competitive (output) layer is implemented, which allows not only simplifies the structure of the neural network classifier but also speeds up the classification process almost 2 times, which is confirmed by the simulation results.