On Model of Recurrent Neural Network on a Time Scale: Exponential Convergence and Stability Research.

The majority of the results on modeling recurrent neural networks (RNNs) are obtained using delayed differential equations, which imply continuous time representation. On the other hand, these models must be discrete in time, given their practical implementation in computer systems, requiring their versatile utilization across arbitrary time scales. Hence, the goal of this research is to model and investigate the architecture design of a delayed RNN using delayed differential equations on a time scale. Internal memory can be utilized to describe the calculation of the future states using discrete and distributed delays, which is a representation of the deep learning architecture for artificial RNNs. We focus on qualitative behavior and stability study of the system. Special attention is paid to taking into account the effect of the time-scale parameters on neural network dynamics. Here, we delve into the exploration of exponential stability in RNN models on a time scale that incorporates multiple discrete and distributed delays. Two approaches for constructing exponential estimates, including the Hilger and the usual exponential functions, are considered and compared. The Lyapunov-Krasovskii (L-K) functional method is employed to study stability on a time scale in both cases. The established stability criteria, resulting in an exponential-like estimate, utilizes a tuple of positive definite matrices, decay rate, and graininess of the time scale. The models of RNNs for the two-neuron network with four discrete and distributed delays, as well as the ring lattice delayed network of seven identical neurons, are numerically investigated. The results indicate how the time scale (graininess) and model characteristics (weights) influence the qualitative behavior, leading to a transition from stable focus to quasiperiodic limit cycles.

Ensemble of RNN Classifiers for Activity Detection Using a Smartphone and Supporting Nodes.

Nowadays, sensor-equipped mobile devices allow us to detect basic daily activities accurately. However, the accuracy of the existing activity recognition methods decreases rapidly if the set of activities is extended and includes training routines, such as squats, jumps, or arm swings. Thus, this paper proposes a model of a personal area network with a smartphone (as a main node) and supporting sensor nodes that deliver additional data to increase activity-recognition accuracy. The introduced personal area sensor network takes advantage of the information from multiple sensor nodes attached to different parts of the human body. In this scheme, nodes process their sensor readings locally with the use of recurrent neural networks (RNNs) to categorize the activities. Then, the main node collects results from supporting sensor nodes and performs a final activity recognition run based on a weighted voting procedure. In order to save energy and extend the network's lifetime, sensor nodes report their local results only for specific types of recognized activity. The presented method was evaluated during experiments with sensor nodes attached to the waist, chest, leg, and arm. The results obtained for a set of eight activities show that the proposed approach achieves higher recognition accuracy when compared with the existing methods. Based on the experimental results, the optimal configuration of the sensor nodes was determined to maximize the activity-recognition accuracy and reduce the number of transmissions from supporting sensor nodes.

Classifier-Based Data Transmission Reduction in Wearable Sensor Network for Human Activity Monitoring.

The recent development of wireless wearable sensor networks offers a spectrum of new applications in fields of healthcare, medicine, activity monitoring, sport, safety, human-machine interfacing, and beyond. Successful use of this technology depends on lifetime of the battery-powered sensor nodes. This paper presents a new method for extending the lifetime of the wearable sensor networks by avoiding unnecessary data transmissions. The introduced method is based on embedded classifiers that allow sensor nodes to decide if current sensor readings have to be transmitted to cluster head or not. In order to train the classifiers, a procedure was elaborated, which takes into account the impact of data selection on accuracy of a recognition system. This approach was implemented in a prototype of wearable sensor network for human activity monitoring. Real-world experiments were conducted to evaluate the new method in terms of network lifetime, energy consumption, and accuracy of human activity recognition. Results of the experimental evaluation have confirmed that, the proposed method enables significant prolongation of the network lifetime, while preserving high accuracy of the activity recognition. The experiments have also revealed advantages of the method in comparison with state-of-the-art algorithms for data transmission reduction.

Distribution Transformer Parameters Detection Based on Low-Frequency Noise, Machine Learning Methods, and Evolutionary Algorithm.

The paper proposes a method of automatic detection of parameters of a distribution transformer (model, type, and power) from a distance, based on its low-frequency noise spectra. The spectra are registered by sensors and processed by a method based on evolutionary algorithms and machine learning. The method, as input data, uses the frequency spectra of sound pressure levels generated during operation by transformers in the real environment. The model also uses the background characteristic to take under consideration the changing working conditions of the transformers. The method searches for frequency intervals and its resolution using both a classic genetic algorithm and particle swarm optimization. The interval selection was verified using five state-of-the-art machine learning algorithms. The research was conducted on 16 different distribution transformers. As a result, a method was proposed that allows the detection of a specific transformer model, its type, and its power with an accuracy greater than 84%, 99%, and 87%, respectively. The proposed optimization process using the genetic algorithm increased the accuracy by up to 5%, at the same time reducing the input data set significantly (from 80% up to 98%). The machine learning algorithms were selected, which were proven efficient for this task.

Classification of Low Frequency Signals Emitted by Power Transformers Using Sensors and Machine Learning Methods.

This paper proposes a method of automatically detecting and classifying low frequency noise generated by power transformers using sensors and dedicated machine learning algorithms. The method applies the frequency spectra of sound pressure levels generated during operation by transformers in a real environment. The spectra frequency interval and its resolution are automatically optimized for the selected machine learning algorithm. Various machine learning algorithms, optimization techniques, and transformer types were researched: two indoor type transformers from Schneider Electric and two overhead type transformers manufactured by ABB. As a result, a method was proposed that provides a way in which inspections of working transformers (from background) and their type can be performed with an accuracy of over 97%, based on the generated low-frequency noise. The application of the proposed preprocessing stage increased the accuracy of this method by 10%. Additionally, machine learning algorithms were selected which offer robust solutions (with the highest accuracy) for noise classification.

A Neuroevolutionary Approach to Controlling Traffic Signals Based on Data from Sensor Network.

The paper introduces an artificial neural network ensemble for decentralized control of traffic signals based on data from sensor network. According to the decentralized approach, traffic signals at each intersection are controlled independently using real-time data obtained from sensor nodes installed along traffic lanes. In the proposed ensemble, a neural network, which reflects design of signalized intersection, is combined with fully connected neural networks to enable evaluation of signal group priorities. Based on the evaluated priorities, control decisions are taken about switching traffic signals. A neuroevolution strategy is used to optimize configuration of the introduced neural network ensemble. The proposed solution was compared against state-of-the-art decentralized traffic control algorithms during extensive simulation experiments. The experiments confirmed that the proposed solution provides better results in terms of reduced vehicle delay, shorter travel time, and increased average velocity of vehicles.

A Survey and Comparison of Low-Cost Sensing Technologies for Road Traffic Monitoring.

This paper reviews low-cost vehicle and pedestrian detection methods and compares their accuracy. The main goal of this survey is to summarize the progress achieved to date and to help identify the sensing technologies that provide high detection accuracy and meet requirements related to cost and ease of installation. Special attention is paid to wireless battery-powered detectors of small dimensions that can be quickly and effortlessly installed alongside traffic lanes (on the side of a road or on a curb) without any additional supporting structures. The comparison of detection methods presented in this paper is based on results of experiments that were conducted with a variety of sensors in a wide range of configurations. During experiments various sensor sets were analyzed. It was shown that the detection accuracy can be significantly improved by fusing data from appropriately selected set of sensors. The experimental results reveal that accurate vehicle detection can be achieved by using sets of passive sensors. Application of active sensors was necessary to obtain satisfactory results in case of pedestrian detection.

Justified granulation aided noninvasive liver fibrosis classification system.

BACKGROUND: According to the World Health Organization 130-150 million (according to WHO) of people globally are chronically infected with hepatitis C virus. The virus is responsible for chronic hepatitis that ultimately may cause liver cirrhosis and death. The disease is progressive, however antiviral treatment may slow down or stop its development. Therefore, it is important to estimate the severity of liver fibrosis for diagnostic, therapeutic and prognostic purposes. Liver biopsy provides a high accuracy diagnosis, however it is painful and invasive procedure. Recently, we witness an outburst of non-invasive tests (biological and physical ones) aiming to define severity of liver fibrosis, but commonly used FibroTest®, according to an independent research, in some cases may have accuracy lower than 50 %. In this paper a data mining and classification technique is proposed to determine the stage of liver fibrosis using easily accessible laboratory data. METHODS: Research was carried out on archival records of routine laboratory blood tests (morphology, coagulation, biochemistry, protein electrophoresis) and histopathology records of liver biopsy as a reference value. As a result, the granular model was proposed, that contains a series of intervals representing influence of separate blood attributes on liver fibrosis stage. The model determines final diagnosis for a patient using aggregation method and voting procedure. The proposed solution is robust to missing or corrupted data. RESULTS: The results were obtained on data from 290 patients with hepatitis C virus collected over 6 years. The model has been validated using training and test data. The overall accuracy of the solution is equal to 67.9 %. The intermediate liver fibrosis stages are hard to distinguish, due to effectiveness of biopsy itself. Additionally, the method was verified against dataset obtained from 365 patients with liver disease of various etiologies. The model proved to be robust to new data. What is worth mentioning, the error rate in misclassification of the first stage and the last stage is below 6.5 % for all analyzed datasets. CONCLUSIONS: The proposed system supports the physician and defines the stage of liver fibrosis in chronic hepatitis C. The biggest advantage of the solution is a human-centric approach using intervals, which can be verified by a specialist, before giving the final decision. Moreover, it is robust to missing data. The system can be used as a powerful support tool for diagnosis in real treatment.

Microwave Saturation of Complex EPR Spectra and Free Radicals of Burnt Skin Treated with Apitherapeutic Agent.

The effect of microwave power on the complex electron paramagnetic resonance spectra of the burn matrix after the therapy with propolis was examined. The spectra were measured with microwaves in the range of 2.2-79 mW. Three groups of free radicals were found in the damaged skin samples. Their spectral lines evolve differently with the microwave power. In order to detect these free radical groups, the lineshape of the spectra was numerically analysed. The spectra were a superposition of three component lines. The best fit was obtained for the deconvolution of the experimental spectra into one Gauss and two Lorentz lines. The microwave power changes also the lineshape of the spectra of thermally injured skin treated with the conventional agent-silver sulphadiazine. The spectral changes were different for propolis and for silver sulphadiazine. The number of individual groups of free radicals in the wound bed after implementation of these two substances is not equal. It may be explained by a higher activity of propolis than of silver sulphadiazine as therapeutic agents.

Application of electron paramagnetic resonance spectroscopy to comparative examination of different groups of free radicals in thermal injuries treated with propolis and silver sulphadiazine.

Different groups of free radicals expressed in burn wounds treated with propolis and silver sulphadiazine were examined. The thermal effect forms major types of free radicals in a wound because of the breaking of chemical bonds. Free radicals, located in the heated skin, were tested after 21 days of treating by these two substances. The aim of this work was to find the method for determination of types and concentrations of different groups of free radicals in wound after high temperature impact during burning. The effects of the therapy by propolis and silver sulphadiazine on free radicals were studied. Since the chemical methods of free radicals studies are destructive, the usefulness of the electron paramagnetic resonance spectroscopy was tested in this work. The electron paramagnetic resonance spectra measured with the microwave power of 2.2 mW were numerically fitted by theoretical curves of Gaussian and Lorentzian shapes. The experimental electron paramagnetic resonance spectra of tissue samples are best fitted by the sum of one Gauss and two Lorentz lines. An innovatory numerical procedure of spectroscopic skin analysis was presented. It is very useful in the alternative medicine studies.