In the article were researched the principles of building systems for observation and recognition of objects. Also we have given the classification of human faces recognition methods. Authors have analized the features of operetion for the progressive calibration network (PCN) for human face recognition. And finally has been created and tested the developed face recognition algorithm as the realized software system.
The possibility of using a neural network to implement a system of avoidance of obstacles on the road has been investigated. The algorithms based on which such a system can work has been reviewed, also the principle of learning of the neural network has been considered. In order to implement investigation the simulator based on Unity and ML Agents has been developed. Using simulator the efficiency of education and this neural network in different configurations has been investigated.
A model of parallel sorting neural network of discrete-time is presented. The model is described by a system of differential equations and by step functions. The network has high speed, any finite resolution of input data and it can process unknown input data of finite values located in arbitrary finite range. The network is characterized by moderate computational complexity and complexity of hardware implementation. The results of computer simulation illustrating the efficiency of the network are provided.
The article describes problems of determining the type and automatic sorting of household waste using mobile computing devices. All of the required hardware and partially software, required for implementation of this service, are already present in modern smartphones.
Object. Development of a method for predicting a two-three dimensional velocity model of a medium by using a shear wave. Complexly structured geological medium is studied on the basis of geophysical and geological data using an artificial neural network. Method. It provides the construction and use of medium models according to geophysical well logging data and other terrestrial geophysical methods. In contrast to existing methods, the proposed method also uses additional data on the medium.
A hardware implementation design of parallelized fuzzy Adaptive Resonance Theory neural network is described and simulated. Parallel category choice and resonance are implemented in the network. Continuous-time and discrete-time winner-take-all neural circuits identifying the largest of M inputs are used as the winner-take-all units. The continuous-time circuit is described by a state equation with a discontinuous right-hand side. The discrete-time counterpart is governed by a difference equation.
Synthesis methods of controllers based on the use of frequency characteristics or root hodographs are considered classic or traditional. Frequency methods are available in practical applications, and most control systems are designed based on various modifications to these methods. A distinctive feature of these methods is the so-called robustness, which means that the characteristics of a closed system are insensitive to the minor errors of the model of the real system.
This paper shows a new type of artificial neural network with dynamic oscillatory neurons that have natural frequencies. Artificial neural network in the mode of information resonance implements a new method of recognition of multispectral images. The constructed neural network will recognize the input spectral images with the amplitude of the non-stationary signal commensurate with the amplitude of the noise signal, due to the resonance effect in nonlinear oscillatory neurons.
The current article describes the results of the study of the neural networks temperature prediction error dependence on measurement errors, which are random, nonlinear and multiplicative errors. It is noted applicability of the architecture of neural network for temperature prediction. The formula of temperature step response for ideal sensor is given.
Current article considers the results of the study of air and water flow temperature prediction error on the number of inputs in neural network. Authors guide the architecture of neural network for temperature prediction. The formula of temperature step response for real sensor is given. Also, the method for calculating the time constants for the temperature step response formula using real measurement data is considered.