Neural networks in frequency bin detection

The Sixth International Symposium on Neural Networks (ISNN 2009)

Background. The disadvantage of spectrally efficient signals with frequency multiplexing is the occurrence of intersymbol interference, which is further aggravated when these signals propagate in frequency selective channels.
 Aim. The possibility and effectiveness of using neural network approaches for channel equalization and signal detection in communication systems using SEFDM signals has been assessed.
 Methods. A receiver structure for SEFDM systems based on a deep complex-valued convolutional neural network is proposed, which allows recovering bits from the temporal representation of the signal without using the fractional Fourier transform and inverting the cross-correlation matrix between frequency subcarriers. A two-stage network training scheme has been developed. Based on simulation modeling, a comparative analysis of the noise immunity of SEFDM systems was carried out both in a channel with white Gaussian noise and in a channel with Rayleigh fading, using classical and neural network receivers.
 Results. It is shown that there is no loss of noise immunity in channels with additive white Gaussian noise and an increase in noise immunity of the system up to 2 dB in the channel specified by the extended automotive model (3GPP-EVA).
 Conclusion. The effectiveness of using deep neural complex-valued convolutional networks as receivers for spectrally efficient communication systems, as well as their advantage over classical ones, is shown.
 Keywords – SEFDM; deep complex-valued convolutional neural network; turbocoding.

Low sidelobe levels may be a crucial factor in the success of achieving high gain using linear receiving arrays, since the level of the sidelobes determines the degree to which noisy nearby sources can be prohibited from interfering with the detection of a weak source on a different azimuth. In real world experiments at sea, one frequently operates with an array which has not only errors in amplitude and phase between channels but also dead elements. The effect of random errors in both amplitude and phase is to yield a sidelobe level of the average beam pattern which is the sum of the error-free pattern and an additional level due to the random errors. The sidelobe level due to the random errors depends upon the details of the amplitude shading function, the distribution function from which the random errors are drawn, and increases with the mean square value of the errors. Thus, given rms values of amplitude and phase errors, there is a practical limit to which the sidelobes can be reduced by amplitude shading the array. The effect of element failure on the sidelobe level can be drastic and depends on the sum of the amplitude weights of the dead elements. Thus, a failure of a highly-weighted element will have a greater effect on the sidelobe level than the failure of two or three whose combined weights are less. [Work supported by NAVMAT.]

We compare the generalization characteristics of complex-valued and real-valued feedforward neural networks. We assume a task of function approximation with phase shift and/or amplitude change in signals having various coherence. Experiments demonstrate that complex-valued neural networks show smaller generalization error than real-valued networks having doubled input and output neurons in particular when the signals have high coherence, that is, high degree of wave nature. We also investigate the relationship between amplitude and phase errors. It is found in real-valued networks that abrupt change in amplitude is often accompanied by steep change in phase, which is a consequence of local minima in real-valued supervised learning.

The amplitude and phase errors deteriorate the low-sidelobe level characteristic of phased array antenna, which degrades radar performance seriously. Therefore an improved calibration method of amplitude and phase errors based on diagonal loading is proposed to address the problem of amplitude and phase errors in this paper. Initially, the mathematical models of amplitude and phase errors are developed and the impact of amplitude and phase errors is analyzed. It is found that although the interference can be suppressed when there is amplitude and phase errors in the array, the low-sidelobe level characteristic of adaptive antenna is badly deteriorated. An improved calibration method of amplitude and phase errors based on diagonal loading is proposed. Compared with traditional calibration method, proposed method can reduce the sidelobe level efficiently. At last simulations were carried out to verify the effectiveness of proposed method.

This paper explores the use and applications of neural networks in the construction of a text markup program. This research paper describes various tasks that require textual data analysis and discusses the problems, issues, and solutions that accompany them. Interest in neural networks has increased in recent years, and they are find­ing applications in a wide variety of fields, such as business, medicine, engineering, geology, and physics. Neural networks have made great strides in forecasting, planning, and management. There are several reasons for this situation. Neural networks are very powerful modeling systems capable of creating complex dependencies. Neural networks can be widely used in areas such as text/speech recognition, semantic search, decision support/expert systems, inventory forecasting, data storage systems and content analysis. The object of the work is the process of functioning of neural networks and an algorithm for text markup recognition. The purpose of the scientific paper is the application of neural networks in the construction of the program for the markup of the text. In order to achieve the goal, the following tasks were put forward: a) study existing neural networks, b) choose a neural network to cre­ate a model and study its structure, c) convert input data to feed it into a neural network model. Representation and analysis of symbolic structures in neural networks seems to be an interesting and useful direction in neural network theory. In this paper, we have reviewed some neural network architectures that may merit further consideration in these circumstances. In what follows, we will focus on specific experiments in this area. Thus, this paper outlines a problem area for further research and testing.

Optically pumped magnetometer (OPM) has gained popularity as a viable alternative to the superconducting quantum interference device (SQUID) for biomagnetic measurements. However, the presence of amplitude and phase errors in the OPM can degrade the signal quality of, e.g., magnetoencephalography (MEG) signals. This study presents a dual-PI closed-loop control (DPCC) method aimed at suppressing amplitude and phase errors of the OPM when operated in the spin-exchange relaxation-free (SERF) regime, thereby enhancing MEG localization accuracy. The main sources of the amplitude and phase errors in the OPM are identified through a theoretical analysis of the inertial element characteristics and static residual magnetic fields. The experimental results demonstrate that the utilization of DPCC for the OPM reduces the amplitude and phase errors in static MEG measurements by 94.2% and 95.6%, respectively. At a residual magnetic field of 6.5 nT, the amplitude and phase errors are further suppressed by 97.8% and 87.2%, respectively. Furthermore, the fluctuation of the residual magnetic field is reduced by a factor of 25, significantly enhancing the robustness of the OPM. This straightforward and effective arrangement of the OPM holds promise in the field of biomagnetism and is expected to enhance the localization and calibration accuracy of OPM-based MEG (OPM-MEG) systems.

On the quadrature accuracy of in-phase coupled quadrature LC oscillator

Objects placed in mammograms and their features have high diagnostic value. The method of extraction and calculation parameters of the images is a key element in the whole process of assessment of individual medical cases. This applies both to detect the location of microcalcifications, as well as the accurate assessment of the degree of malignancy [1, 2, 3, 4, 8, 9]. The analysis of the literature shows that in many cases, to assess them as neural classifiers used neural networks unidirectional scholars algorithm back propagation of error. The large popularity of this type of network stems from the rich knowledge about them supported a wide range of applications [11, 12, 13, 17, 18]. This does not mean that this type of network is the best tool for the identification of cancer. In order to verify the correctness of the research, it was decided to compare the two types of neural networks: one-way multi-layer network MLBP [17, 18] and a network of independently building the architecture i.e. Fahlmana network [5, 17]. Both networks were used for the same task, the task of verification and classification of medical data based on the feature vectors. The study attempted to assess the mammograms for detecting the locus and the evaluation grade of microcalcifications [19, 23]. According to the stated objective, the structure of the neural networks were designed so that it was possible to indicate location and performance evaluation grade localized potential microcalcifications. Way of describing mammogram—a set of characteristics and size of the database are the factors that were taken as those that have a key impact on the quality of the diagnostic computer tools [21, 22, 23].

The purpose of the studyis to examine the theory and practice ofusing neural networks in education, to develop the concept of aneural network adaptive learning environment, and to implementa neural network model of one of the subsystems of this environment (using a model example of creating an adaptive educationaltrajectory).Materials and methods.The study includes a review of bibliographicsources on the application of neural networks in the field of education.It also includes modeling the structure of a neural network adaptivelearning environment. The PyTorch library was used to programmatically implement the neural network.Results.The prospects for the use of neural networks in the field ofeducation are considered, including various tasks of recognition,diagnostics, classification, clustering, forecasting, optimization,etc. A structural model of an adaptive learning NeuroSmartenvironment is created. This environment includes a number ofsubsystems for solving the following tasks: biometric identificationof the student; determining starting level; choosing an adaptivelearning path; determining the current level of competency development; automated verification of students’ work using recognitiontechnology; analysis of the final result of training; monitoringinformation security incidents in an electronic course, etc. Inorder to study the possibilities and problems of applying neuralnetwork models to the tasks of constructing student adaptivelearning trajectories, a model example of a neural network wascreated. This example illustrates the possibility of using a neural network to select further nodes of the educational trajectorybased on the available data on the current learning parametersin an electronic course. To implement the neural network, thePyTorch deep learning library and Pandas library modules wereused. SGD, Adam, Rprop were used as an optimizer to performgradient descent steps. For each of the optimizers, a computerstudy of the stability of the neural network was carried out byvarying the following parameters: the learning rate coefficient,the number of neurons in the hidden layer, and the number oftraining epochs.Conclusion.It can be assumed that the next stage in the evolutionof the use of neural network technologies in the field of educationwill be their integration into complex multi-component Smart systemscapable of automatically supporting student learning at all stagesof the implementation of their personal educational trajectory. Obviously, the practical implementation of complex neural networksystems of this level is a very difficult task and can be solved so faronly at the level of individual subsystems. There are a number of problems associated with computer simulation of the educationalenvironment based on neural network models: the question of theoptimal structure of the neural network (for example, the numberof neurons and layers in the network) has not been studied enough,there are no clear criteria for the optimality of the adaptive educational trajectory. Nevertheless, it should be noted that the task ofdeveloping new forms and technologies of personalized e-learning isin great demand, which makes modeling based on neural networksespecially relevant.

Application of neural networks in detection of abnormal brain perfusion regions

An 8-elements Ku-band phased-array transmitter with built-in-self-test (BIST) capability is presented. A fully symmetrical coupler feed line is used to improve the accuracy of the BIST and calibration procedure. 5-bit phase shifters and 4-bit attenuators with less phase and amplitude errors are designed to control the beam direction and side lobes. The RMS phase and amplitude errors of the phase shifters are less than 3.6° and 0.8 dB, respectively. The RMS amplitude and phase errors of the attenuators are less than 0.94 dB and 3.2°. The measured single path power gain including the power divider is 11.9 dB. The output referred PldB is 13 dBm for each channel. ESD protection and temperature sensors are integrated. The chip is fabricated in 180 nm bulk CMOS technology. Compact design is adopted to minimize the size of each channel, which is suitable for the larze-scale phased-array system. The size of the whole chip is $\mathbf{8.5x3 mm}^{\mathbf{2}}$ , and that of each channel is $\mathbf{3.8x0.5 mm}^{\mathbf{2}}$ .

Application of neural networks in electrical power systems have demonstrated that this powerful tool can be employed as an alternative method for solving problems accurately and efficiently. In this paper, an artificial neural networks approach is presented for the detection, classification and isolation (location) of transmission line faults occurring in the system of transmission line. The main aim is to implement a complete scheme for distance protection of a transmission line system. In order to perform this goal, the distance protection task is subdivided into different neural network for fault detection, fault identification (classification) as well as fault location in different zones. 110 KV transmission line system is used to create training and testing data for the neural network by mat-lab program. The result show that the neural network is a powerful analysis tool for providing faults information that is more reliable and accurate than provided by the traditional relay.

A solo attack may cause a big loss in computer and network systems, its prevention is, therefore, very inevitable. Precise detection is very important to prevent such losses. Such detection is a pivotal part of any security tools like intrusion detection system, intrusion prevention system, and firewalls etc. Therefore, an approach is provided in this paper to analyze denial of service attack by using a supervised neural network. The methodology used sampled data from Kddcup99 dataset, an attack database that is a standard for judgment of attack detection tools. The system uses multiple layered perceptron architecture and resilient backpropagation for its training and testing. The developed system is then applied to denial of service attacks. Moreover, its performance is also compared to other neural network approaches which results more accuracy and precision in detection rate.

At present, our life is almost inseparable from the network, the network provides a lot of convenience for our life. However, a variety of network security incidents occur very frequently. In recent years, with the continuous development of neural network technology, more and more researchers have applied neural network to intrusion detection, which has developed into a new research direction in intrusion detection. As long as the neural network is provided with input data including network data packets, through the process of self-learning, the neural network can separate abnormal data features and effectively detect abnormal data. Therefore, the article innovatively proposes an intrusion detection method based on deep convolutional neural networks (CNN), which is used to test on public data sets. The results show that the model has a higher accuracy rate and a lower false negative rate than traditional intrusion detection methods.