Pattern Recognition Module Research Articles

Intelligent water treatment management system for swimming pools

The article presents the assessment methodology for anthropogenic water pollution in swimming pools, based on intelligent video monitoring of people physical activity, as well as a model for correlation data analysis and a regression model for the main parameters of chemical water contamination and doses of introduced reagents. The ultimate goal of the project is to create an intelligent water treatment management system for swimming pools. It is planned to use YOLOv4 neural networks to implement an intelligent pattern recognition module (swimmers in the pool) in the data stream coming from video cameras in real time. This innovative system will make it possible to automatically make managerial and technical decisions to determine the required amount of introduced reagents at various time intervals depending on the actual amount of organic contaminants, as well as other factors determined by the results of the correlation-regression analysis of the data. A review of information resources has shown that such a methodology does not exist both in Russia and abroad.

Deep multi-scale adversarial network with attention: A novel domain adaptation method for intelligent fault diagnosis

Data driven-based intelligent fault diagnosis methods, as a promising approach, have been widely employed in the health management and maintenance decision of rotating machinery. However, the domain shift phenomenon caused by internal and external interference inevitably exists in practical application scenarios, which significantly deteriorates the performances of the intelligent diagnosis model. And the preparation of label information in real complex scenes is usually time-consuming and expensive. To overcome these challenges, a novel unsupervised domain adaptation framework named deep multi-scale adversarial network with attention (MSANA) is introduced for machinery fault diagnosis. It is established based on two main components, one is the shared feature generator, which is constructed by two novel multi-scale modules with attention mechanism, and the other part is a fault pattern recognition module composed of two differentiated discriminators. While the multi-scale module is used to obtain rich features through different internal perceptual scales, the attention mechanism determines the weights of different scales, which promotes the dynamic adjustment performance and adaptive ability of the model. Then, decision boundary assisted adversarial learning strategy is employed to eliminate domain distribution differences and obtain domain-invariant features. A total of ten rolling bearing-based transfer scenarios and six gearbox-based transfer scenarios are adopted to evaluate the transferability of the proposed MSANA model, and the cross-domain transfer results show that it has superior transferability and stability.

Adaptive Control Strategy Extracted from Dynamic Programming and Combined with Driving Pattern Recognition for SPHEB

An appropriate control strategy can play an important role in further improving the fuel economy performance of hybrid electric vehicle (HEV). This research developed a novel adaptive control strategy to achieve optimal power distribution for a series-parallel hybrid electric bus (SPHEB) to adapt driving pattern instantaneously. First, a methodology of extracting mode transition control and power distribution strategy from dynamic programming (DP) solution is proposed for the development of the hierarchical energy management strategy. A SPHEB energy management problem under the Chinese typical bus driving schedule at urban district (CTBDS_UD) is investigated as a case study. Second, an approach of driving pattern recognition (DPR) module is developed. For adaptive learning, four typical driving patterns are selected as the database of driving condition and using the extraction method described above to acquire optimal control strategies for four driving patterns. Third, a framework of adaptive control strategy has been proposed based on the extracted hierarchical energy management strategy from DP and combined with DPR. Finally, the simulation results demonstrate the proposed adaptive strategy can make power distribution proper adjustments in real time and be capable of improving significantly the fuel efficiency of the SPHEB.

Noise filtering framework for electronic nose signals: An application for beef quality monitoring

Beef is one of the most popular and widely consumed foodstuffs in the world. Nevertheless, it can easily decay if not properly treated during distribution and storage. The consumption of low quality beef causes a serious health hazard. The electronic nose (e-nose) is a rapid and low-cost instrument for beef quality classification. Hence, the development of a mobile e-nose for online meat quality monitoring is appealing. In the last few years, e-noses have been used to classify different grades of beef and to predict the number of the microbial population in beef samples. Several methods are used to deal with these classification and regression problems. Especially in multiclass beef classification and regression, signals contaminated with noise can significantly degrade the performance of the pattern recognition module. Therefore, the presence of internal and external noise in e-nose signals is a major challenge in beef quality monitoring. In this study, a noise filtering framework based on a fine-tuned discrete wavelet transform (DWT) was developed to handle noisy signals generated by an e-nose sensor array. To the best of our knowledge this is the first time the problem of e-nose signal noise in beef quality classification is tackled. The proposed framework was integrated and tested on several machine learning algorithms that were used in previous studies, i.e. k-nearest neighbor (k-NN), support vector machine (SVM), quadratic discriminant analysis (QDA), artificial neural network (ANN), and adaptive neuro fuzzy inference system (ANFIS). Furthermore, the effect of noise filtering was investigated in the classification with two, three, and four classes of beef. The effect of noise filtering was also observed in regression tasks to predict the size of microbial population in beef samples. The experimental results showed that the proposed framework provides a significant improvement in multiclass classification and regression tasks.

Graphene Foam Chemical Sensor System Based on Principal Component Analysis and Backpropagation Neural Network

A kind of graphene foam chemical sensor (GFCS) system based on the principal component analysis (PCA) and backpropagation neural network (BPNN) was presented in this paper. Compared with conventional chemical sensors, the GFCS could discriminate various chemical molecules with selectivity without surface modification. The GFCS system consisted of an unmodified graphene foam chemical sensor, an electrical resistance time domain detection system (ERTDS), and a pattern recognition module. The GFCS has been validated via several chemical molecules discrimination including chloroform, acetone, and ether. The experimental results showed that the discrimination accuracy for each molecule exceeded 97% and a single measurement can be achieved in ten minutes. This work may have presented a new strategy for research and application for graphene chemical sensors.

Reproducibility and Prognostic Potential of Ki-67 Proliferation Index when Comparing Digital-Image Analysis with Standard Semi-Quantitative Evaluation in Breast Cancer

In this study, the reproducibility of Ki-67 proliferation index (KIPI) was investigated by comparing the semi-quantitative (SQ) results of three assessors with those of digital image-analysis (DIA) methods. The prognostic significance of the two approaches was also correlated with clinical outcome. Tissue microarrays of duplicate 2mm cores were constructed from representative areas of formalin-fixed and paraffin-embedded tumor blocks of 347 breast cancer patients. SQ evaluation of Ki-67 (MIB1 clone) immunostained slides was performed independently by three pathologists. DIA was completed using a fully automated histological pattern and cell recognition module for KIPI detection (DIA-1) and an adjustable module (DIA-2) with the possibility of manual corrections. To compare SQ and DIA evaluations intra-class correlation (ICC) and concordance correlation coefficients (CCC) were determined. The three SQ evaluations demonstrated a remarkable ICC (0.853). Significant difference and poor concordance occurred between SQ-1 and SQ-2 as well as between SQ-1 and SQ-3 (p≤0.001, CCC≤0.827 for both comparisons). Thus, the reference KIPI value (SQ-RV) was generated from the mean values of SQ-2 and SQ-3. SQ-RV and DIA-2 results showed substantial concordance (CCC=0.963, at p=0.754), while SQ-RV and DIA-1 values differed (p≤0.001) at only moderate concordance (CCC=0.906). In multivariate analysis, lymph node status and SQ-2 assessment were significantly associated with clinical outcome (p≤0.012 for both comparisons). Our results confirm that KIPI is a significant prognostic marker in breast cancer, which can be can be reliably reproduced by using an adjustable DIA-2 image analysis module.

Learning Algorithm for the Recursive Pattern Recognition Model

ABSTRACTIn this work, we incorporate a learning algorithm to the recursive pattern recognition model, based on the systematic functioning of the human neocortex presented in previous works. This algorithm has two mechanisms: the first, called Aprendizaje_nuevo, is used to learn new patterns and creates a new pattern recognition module in the model. The other, called Aprendizaje_por_refuerzo, is used to reinforce a pattern and adapts the module that represents the pattern to the changes in it. The algorithm is tested in various contexts (text and images) to analyze its capacities of learning and of recognition of the model.

An End-User Platform for FPGA-Based Design and Rapid Prototyping of Feedforward Artificial Neural Networks With On-Chip Backpropagation Learning

The hardware implementation of an artificial neural network (ANN) using field-programmable gate arrays (FPGAs) is a research field that has attracted much interest and attention. With the developments made, the programmer is now forced to face various challenges, such as the need to master various complex hardware–software development platforms, hardware description languages, and advanced ANN knowledge. Moreover, such an implementation is very time consuming. To address these challenges, this paper presents a novel neural design methodology using a holistic modeling approach. Based on the end-user programming concept, the presented solution empowers end users by means of abstracting the low-level hardware functionalities, streamlining the FPGA design process and supporting rapid ANN prototyping. A case study of an ANN as a pattern recognition module of an artificial olfaction system trained to identify four coffee brands is presented. The recognition rate versus training data features and data representation was analyzed extensively.

A Low-Cost Real-Time Research Platform for EMG Pattern Recognition-Based Prosthetic Hand

The focus of this paper is the development of a low-cost research platform for a surface electromyogram (EMG)-based prosthetic hand control to evaluate various pattern recognition techniques and to study the real-time implementation. This comprehensible research platform may enlighten the biomedical research in developing countries for analyzing and evaluating the surface EMG signals. Major challenges in this work were as follows: the design and development of an EMG signal conditioning module, a pattern recognition module, and a prosthetic hand at low cost. Besides, EMG pattern recognition techniques were evaluated for identifying six hand motions in offline with signals acquired from ten healthy subjects and two transradial amputees. Features calculated from EMG signals were grouped into six ensembles to apprehend the vitality of the ensemble in classifiers namely simple logistic regression (SLR), J48 algorithm for decision tree, logistic model tree, neural network, linear discriminant analysis, and support vector machine. The classification performance was also evaluated with the prolonged EMG data recorded on a day at every 1 h interval to study the robustness of the classifier. The results show the average classification accuracy, processing time and memory requirement of the SLR was found to be better and robust with time-domain features consisting of statistical as well as autoregression coefficients. The statistical analysis of variance test also showed that computation time and memory space required for SLR were significantly less compared to the other classifiers. The performance of the classifier was tested in online with transradial amputee for actuation of prosthetic hand for two intended motions with a TMS320F28335 controller. This proposed research platform for evaluation of EMG pattern recognition and real-time implementation has been achieved at a cost of 25 000 Indian rupee (INR).

NextMe: Localization Using Cellular Traces in Internet of Things

The Internet of Things (IoT) opens up tremendous opportunities to location-based industrial applications that leverage both Internet-resident resources and phones’ processing power and sensors to provide location information. Location-based service is one of the vital applications in commercial, economic, and public domains. In this paper, we propose a novel localization scheme called NextMe , which is based on cellular phone traces. We find that the mobile call patterns are strongly correlated with the co-locate patterns. We extract such correlation as social interplay from cellular calls, and use it for location prediction from temporal and spatial perspectives. NextMe consists of data preprocessing, call pattern recognition, and a hybrid predictor. To design the call pattern recognition module, we introduce the notions of critical calls and corresponding patterns. In addition, NextMe does not require that the cell tower addresses should be bounded with concrete coordinates, e.g., global positioning system (GPS) coordinates . We validate NextMe across MIT Reality Mining Dataset, involving 500 000 h of continuous behavior information and 112 508 cellular calls. Experimental results show that NextMe achieves fine-grained prediction accuracy at cell tower level in the forthcoming 1–6 h with 12% accuracy enhancement averagely from cellular calls.

Lip Motion Pattern Recognition for Indonesian Syllable Pronunciation Utilizing Hidden Markov Model Method

A speech therapeutic tool had been developed to help Indonesian deaf kids learn how to pronounce words correctly. The applied technique utilized lip movement frames captured by a camera and inputted them in to a pattern recognition module which can differentiate between different vowel phonemes pronunciation in Indonesian language. In this paper, we used one dimensional Hidden Markov Model (HMM) method for pattern recognition module. The feature used for the training and test data were composed of six key-points of 20 sequential frames representing certain phonemes. Seventeen Indonesian phonemes were chosen from the words usually used by deaf kid special school teachers for speech therapy. The results showed that the recognition rates varied on different phonemes articulation, ie. 78% for bilabial/palatal phonemes and 63% for palatal only phonemes. The condition of the lips also had effect on the result, where female with red lips has 0.77 correlation coefficient, compare to 0.68 for pale lips and 0.38 for male with mustaches.

Classification of Radionuclides on Polyvinyl Toluene Radiation Portal Monitors by a Neural Network Based System§

To detect radiation sources in containers, radiation portal monitors (RPMs) are normally used. Among the various types of RPM, polyvinyl toluene (PVT) monitors are the most widely used, thanks to their lower cost. However, they provide only low resolution when energy spectra are measured. For this reason, it is currently not possible to identify what kind of radiation source is being measured, only that there is some radiation source present, when an alarm is triggered in a PVT monitor. In this case, a second inspection is necessary in order to distinguish between innocent alarms and those due to potentially dangerous radioactive material. This second inspection involves a considerable outlay of money and time at frontiers. The main purpose of this paper is to propose an algorithm which can be applied so as to reduce the number of innocent alarms purely on the basis of the data obtained by the PVT monitors. To this end, an intelligent system based on two stages is proposed: a preprocessing of the energy spectrum and a neural network based pattern recognition module, which classifies the radiation source into one of the standard groups. This intelligent system has been tested with real radiation sources, whose energy spectra were measured with two different PVT detectors, yielding good results.

A real-time, practical sensor fault-tolerant module for robust EMG pattern recognition.

BackgroundUnreliability of surface EMG recordings over time is a challenge for applying the EMG pattern recognition (PR)-controlled prostheses in clinical practice. Our previous study proposed a sensor fault-tolerant module (SFTM) by utilizing redundant information in multiple EMG signals. The SFTM consists of multiple sensor fault detectors and a self-recovery mechanism that can identify anomaly in EMG signals and remove the recordings of the disturbed signals from the input of the pattern classifier to recover the PR performance. While the proposed SFTM has shown great promise, the previous design is impractical. A practical SFTM has to be fast enough, lightweight, automatic, and robust under different conditions with or without disturbances.MethodsThis paper presented a real-time, practical SFTM towards robust EMG PR. A novel fast LDA retraining algorithm and a fully automatic sensor fault detector based on outlier detection were developed, which allowed the SFTM to promptly detect disturbances and recover the PR performance immediately. These components of SFTM were then integrated with the EMG PR module and tested on five able-bodied subjects and a transradial amputee in real-time for classifying multiple hand and wrist motions under different conditions with different disturbance types and levels.ResultsThe proposed fast LDA retraining algorithm significantly shortened the retraining time from nearly 1 s to less than 4 ms when tested on the embedded system prototype, which demonstrated the feasibility of a nearly “zero-delay” SFTM that is imperceptible to the users. The results of the real-time tests suggested that the SFTM was able to handle different types of disturbances investigated in this study and significantly improve the classification performance when one or multiple EMG signals were disturbed. In addition, the SFTM could also maintain the system’s classification performance when there was no disturbance.ConclusionsThis paper presented a real-time, lightweight, and automatic SFTM, which paved the way for reliable and robust EMG PR for prosthesis control.

Discrimination between glioblastoma multiforme and solitary metastasis using morphological features derived from the p:q tensor decomposition of diffusion tensor imaging.

The management and treatment of high-grade glioblastoma multiforme (GBM) and solitary metastasis (MET) are very different and influence the prognosis and subsequent clinical outcomes. In the case of a solitary MET, diagnosis using conventional radiology can be equivocal. Currently, a definitive diagnosis is based on histopathological analysis on a biopsy sample. Here, we present a computerised decision support framework for discrimination between GBM and solitary MET using MRI, which includes: (i) a semi-automatic segmentation method based on diffusion tensor imaging; (ii) two-dimensional morphological feature extraction and selection; and (iii) a pattern recognition module for automated tumour classification. Ground truth was provided by histopathological analysis from pre-treatment stereotactic biopsy or at surgical resection. Our two-dimensional morphological analysis outperforms previous methods with high cross-validation accuracy of 97.9% and area under the receiver operating characteristic curve of 0.975 using a neural networks-based classifier.

Rotor Fault Diagnosis and Simulation Based on LabVIEW

A fault diagnosis system based on LabVIEW is researched to analyze the rotating machinery fault. By developing data acquisition, signal analysis, processing and simulation, pattern recognition and fault diagnosis modules, this system can automatically measure and identify the shaft orbit of a rotor and get fault results. A pattern recognition method on the basis of the invariant moment algorithm is studied with LabVIEW and used to recognize shaft orbit shapes. The experimental results show that the shaft orbit of a rotor can be identified through invariant moment calculation. Because the shape of the shaft orbit of the rotor is close to the ellipse the main fault is imbalance. The fault result is proved to be correct by spectrum analysis. This study is helpful to develop a online fault diagnosis system based on LabVIEW.

A Probabilistic Neural Network-Based Module for Recognition of Objects from their 3-D Images

In this paper, a pattern recognition module that makes use of 3-D images of objects is presented. The proposed module takes advantage of both the generalization capability of neural networks and the possibility of manipulating 3-D images to generate views at different poses of the object that is to be recognized. This allows the construction of a robust 3-D object recognition module that can find use in various applications including military, biomedical and mine detection applications. The paper proposes an efficient training procedure and decision making strategy for the suggested neural network. Sample results of testing the module on 3-D images of several objects are also included along with an insightful discussion of the implications of the results.

Chaos based neural network optimization for concentration estimation of indoor air contaminants by an electronic nose

Electronic nose (E-nose), as an artificial olfactory system, can be used for estimation of gases concentration combined with a pattern recognition module. This paper studies the concentration estimations of indoor contaminants for air quality monitoring in dwellings using chaos based optimization artificial neural network integrated into our self-designed portable E-nose instrument. Back-propagation neural network (BPNN) has been recognized as the common pattern recognition. Considering the local optimal flaw of BPNN, this paper presents a novel chaotic sequence optimization BPNN method for improving the accuracy of E-nose prediction. Further comparison with particle swarm optimization is also employed, and maximum 26.03% and 16.4% prediction error decreased after using chaotic based optimization for formaldehyde and benzene concentration estimation. Experimental results demonstrate the superiority and efficiency of the portable E-nose instrument integrated with artificial neural network optimized by chaotic sequence based optimization algorithms in real-time monitoring of air quality in dwellings.

The Syntactical Pattern Recognition for the Leakage Current of Transmission-Line Insulators

For the current analytical method of the leakage current on the insulator, the online monitoring system could not be used to accurately judge the contamination severity on the surface of insulators, and a novel method, syntactical pattern recognition, was proposed in this paper. The geometric primitives were extracted by piecewise linearization from the leakage current signals. In accordance with the given grammar, these primitives are constituted into a language recognition unit with the name of finite automaton, where after the finite automaton, the leakage current signals were judged to be the corresponding pattern. The syntactical pattern-recognition module was embed into the “Acqs” software, and a series of experiments for validating this method's performance were made under the following three different conditions: 1) the leakage current of polluted insulators under the different humidity degrees; 2) the leakage current under the different voltage values with moderate humidity; and 3) the polarity effect. The results showed that the structure information of the leakage current could be roundly obtained by the syntactical pattern-recognition method. Consequently, the surface discharge characteristics of insulators could be deeply grasped and the polarity effect of the leakage current of insulators could be accurately and fully recognized.

Low false alarm target detection and tracking within strong clutters in outdoor infrared videos

Reliable target detection and tracking within strong clutters in outdoor infrared video sequences present great challenges. This is caused by several artificial and natural conditions, such as luminance change resulted from automatic gain adjustment in the IR camera, and other extreme granularity and uncontrolled environmental factors. In some important applications, the targets of interest include vehicles in motion and people in transit. We propose a new integrated solution to address all these issues. The system is composed of the following components: a region masking algorithm to divide a video frame into reliable and unreliable regions, a novel motion pattern recognition module, and an automatic walking-person recognition module. Such a comprehensive technique makes it possible for real-world outdoor infrared surveillance applications, where environmental interferences are uncontrolled and target motion are complex. Extensive experiments were carried out on real-world outdoor infrared videos provided by General Dynamics. In all tested sequences, the proposed algorithm had successfully detected and tracked the targets consistently throughout the life cycle of the targets, even when some targets were seriously blurred or occluded. Our results show that the proposed algorithm is practical, robust, and reliable for low-quality outdoor infrared videos.

A Concept of a Sensor System for Determining Composition of Organic Solvents

This paper presents the results of work on a gas sensor system for determining qualitative and quantitative composition of organic solvents. The system performs the analysis of gaseous mixtures, which are obtained by evaporating liquid solvents and therefore directly represent their composition. The system utilizes gas sensor array, which is built of Taguchi gas sensors (TGS). It provides patterns of measured solvents which are analyzed by pattern recognition module. The module employs multivariate statistical tools like discriminant function analysis (DFA) and partial least squares (PLS) regression. The determination of solvent composition is organized in a hierarchical structure. The concept of the system was experimentally verified using the example of two-component solvents. It was here demonstrated that the proposed system is capable of determining qualitative and quantitative composition of liquid solvents.