Pattern Recognition Tools Research Articles

Multi-Chain Semi-Markov Analysis of Intrapartum Cardiotocography.

Visual assessment of the evolution of fetal heart rate (FHR) and uterine pressure (UP) patterns is the standard of care in the intrapartum period. Unfortunately, this assessment has high levels of intra- and inter-observer variability. This study processed and analyzed FHR and UP patterns using computerized pattern recognition tools. The goal was to evaluate differences in FHR and UP patterns between fetuses with normal outcomes and those who developed hypoxic-ischemic encephalopathy (HIE). For this purpose, we modeled the sequence of FHR patterns and uterine contractions using Multi-Chain Semi-Markov models (MCSMMs). These models estimate the probability of transitioning between FHR or UP patterns and the dwell time of each pattern. Our results showed that in comparison to the control group, the HIE group had: (1) more frequent uterine contractions during the last 12 hours before birth; (2) more frequent FHR decelerations during the last 12 hours before birth; (3) longer decelerations during the last eight hours before birth; and (4) shorter baseline durations during the last five hours before birth. These results demonstrate that the fetuses in the HIE group were subject to a more stressful environment than those in the normal group. Clinical Relevance- Our results revealed statistically significant differences in FHR/UP patterns between the normal and HIE groups in the hours before birth. This indicates that features derived using MCSMMs may be useful in a machine learning framework to detect infants at increased risk of developing HIE allowing preventive interventions.

(Digital Presentation) Data-Driven Prognosis of Battery Failure Detection and Prediction

Although the electric vehicle market is witnessing an unprecedented evolution, the fast adoption of these vehicles requires a more thorough status analysis of the battery performance's functionality and reliability. Due to their rechargeable nature, Lithium-ion batteries (LIBs) operation is subject to different irreversible processes during their charging and discharging cycles and causing capacity fade due to various degradation mechanisms. These processes generally result in battery capacity degradation, which usually results in battery failure, with consequences ranging from loss of operation, reduced capability, downtime, and catastrophic malfunctions. To address the issues mentioned above, numerous studies have been dedicated to proposing proper degradation model mechanisms for improving the reliability and availability of LIBs. However, due to accuracy and computational complexity challenges, most existing remaining useful life (RUL) and health prediction models focus on special degradation effects and ignore the integrated deterioration mechanisms, which generally involve batteries' capacity fade associated with the inadequacy of current health estimation tools. Thus, these shortcomings ultimately echo the need to devise novel tools to identify the dominant criteria negatively affecting battery performance and accurately predict the system's failure. The above challenges eventually necessitate a robust and reliable predictive or prognostic capability for prognostics and health monitoring (PHM) under a complexly hostile working environment. In this context, this investigation aims at proposing a novel data-driven approach called data-driven prognosis (DDP) that estimates the relevant constitutive parameters in situ and captures deviations from the expected degradation dynamics of the LIBs in addition to precise modeling of the degradation and capacity models. This talk will present a new data-driven approach using statistical pattern recognition and machine learning tools to detect batteries' anomalies and failures.

Design of a high-tech system to evaluate fresh strawberry quality-based on the principal component analysis

This study was designed to investigate the differences in the physicochemical properties among twelve strawberry cultivars by using pattern recognition tools, to provide a theoretical basis for quality variation among samples. The data of 14 indicators were subjected to principal component analysis (PCA), descriptive statistical analysis, correlation analysis, and hierarchical cluster analysis, to filter core evaluation factors. Quality evaluation index weights and quality comprehensive evaluation were determined by an analytic hierarchy process. Of the 14 indicators selected as indices of fresh strawberry quality evaluation index, including a value, sugar-acid ratio, firmness, vitamin C, TAC, and TPC. The data were deployed to adjust the multivariate kinetics using Analytic Hierarchy Process (AHP), and the results were compared to those sensory score using sensory personnel. Results showed that the correlation coefficient of sensory scores and AHP comprehensive score is 0.9239. This high correlation coefficient indicates that the use of our mathematical model for strawberry quality evaluation is feasible. The information herein provides a practical strategy for the evaluation of strawberry quality.

Deep clustering variational network for helicopter regime recognition in HUMS

The helicopter health and usage monitoring system is the core system to ensure its operational safety. Flight regime recognition is a key pilot task therein that affects subsequent decision-making. However, the current research on this topic has not aroused wide attention. Taking advantage of deep learning, a powerful pattern recognition tool, we proposed a deep clustering variational network to serve the helicopter regime recognition task. Through explicit feature distribution constraints and clustering loss function, we have made a clearer decision boundary and more significant category differences, thus achieving accurate recognition results. Two case studies show that deep clustering variational network can effectively recognize the regimes by utilizing vibration signals in time between overhaul experiments or online flight parameters.

Liquid State Machine on SpiNNaker for Spatio-Temporal Classification Tasks.

Liquid State Machines (LSMs) are computing reservoirs composed of recurrently connected Spiking Neural Networks which have attracted research interest for their modeling capacity of biological structures and as promising pattern recognition tools suitable for their implementation in neuromorphic processors, benefited from the modest use of computing resources in their training process. However, it has been difficult to optimize LSMs for solving complex tasks such as event-based computer vision and few implementations in large-scale neuromorphic processors have been attempted. In this work, we show that offline-trained LSMs implemented in the SpiNNaker neuromorphic processor are able to classify visual events, achieving state-of-the-art performance in the event-based N-MNIST dataset. The training of the readout layer is performed using a recent adaptation of back-propagation-through-time (BPTT) for SNNs, while the internal weights of the reservoir are kept static. Results show that mapping our LSM from a Deep Learning framework to SpiNNaker does not affect the performance of the classification task. Additionally, we show that weight quantization, which substantially reduces the memory footprint of the LSM, has a small impact on its performance.

Ceruletide and Alpha-1 Antitrypsin as a Novel Combination Therapy for Ischemic Stroke.

Ischemic stroke is a primary cause of morbidity and mortality worldwide. Beyond the approved thrombolytic therapies, there is no effective treatment to mitigate its progression. Drug repositioning combinational therapies are becoming promising approaches to identify new uses of existing drugs to synergically target multiple disease-response mechanisms underlying complex pathologies. Here, we used a systems biology-based approach based on artificial intelligence and pattern recognition tools to generate in silico mathematical models mimicking the ischemic stroke pathology. Combinational treatments were acquired by screening these models with more than 5 million two-by-two combinations of drugs. A drug combination (CA) formed by ceruletide and alpha-1 antitrypsin showing a predicted value of neuroprotection of 92% was evaluated for their synergic neuroprotective effects in a mouse pre-clinical stroke model. The administration of both drugs in combination was safe and effective in reducing by 39.42% the infarct volume 24h after cerebral ischemia. This neuroprotection was not observed when drugs were given individually. Importantly, potential incompatibilities of the drug combination with tPA thrombolysis were discarded in vitro and in vivo by using a mouse thromboembolic stroke model with t-PA-induced reperfusion, revealing an improvement in the forepaw strength 72h after stroke in CA-treated mice. Finally, we identified the predicted mechanisms of action of ceruletide and alpha-1 antitrypsin and we demonstrated that CA modulates EGFR and ANGPT-1 levels in circulation within the acute phase after stroke. In conclusion, we have identified a promising combinational treatment with neuroprotective effects for the treatment of ischemic stroke.

A Comparative Analysis Applied to the Partial Discharges Identification in Dry-Type Transformers by Hall and Acoustic Emission Sensors.

Dry-type insulated transformers stand out for their higher applicability in substations, high-voltage instrumentation systems, and electrical installations. In this machine, the insulation system is constituted of dielectric materials such as epoxy resin and Nomex paper. Some critical issues in the operation of this equipment, such as overload, moisture, or heat, can induce a slow degradation of the physical–chemical properties of the dielectric materials, which can culminate in the total failure of the transformer. However, before the transformer’s shutdown, it is common to detect discharge activity in the insulation system. Based on this issue, this work proposes an experimental and comparative analysis between acoustic emission and Hall-effect sensors, aiming at differentiating discharges in epoxy resin and Nomex paper, materials that constitute the insulation of the dry-type insulated transformers. Two signal processing techniques were studied: traditional frequency analysis and discrete wavelet transform. The objective is to develop signal processing techniques to differentiate each type of discharge since different discharges require different maintenance actions. The results obtained indicate that acoustic emission sensors and Hall sensors are promising in differentiating discharge in epoxy resin and Nomex paper. Furthermore, the pattern recognition tools presented by this work, which associated the wavelet levels energies and the energy of the full signals with the average band and the equivalent bandwidth, were effective to perform feature extraction of power transformer condition.

Classification of surface electromyography and gyroscopic signals of finger gestures acquired by Myo armband using machine learning methods

Gestures of the human hand can be identified through processing of surface electromyography (sEMG) signals. The human hand can perform many gestures via manipulation of the fingers. With correct classification of finger gestures, the mobility of a prosthetic hand can be increased and provide greater functionally. In this study, reliable classification was obtained for sEMG finger data acquired from a Myo armband placed on the lower forearm. In order to improve classification, gyroscopic signals, not previously used in other studies, were investigated in the sEMG finger data. Data was acquired from ten normal subjects using the Myo armband to identify 6 finger gestures: thumb, index finger, middle finger, little finger, ring finger and rest. Participants repeated each gesture thirty times. sEMG signals were preprocessed to extract features. 17 features were used in the feature matrix. By using the sequential forward feature selection method, the highest performance feature set was determined. Support Vector Machine, K-Nearest Neighbor and multilayer artificial neural network were used as classification algorithm. The classification was made using the Classification Learner Application and Neural Network Pattern Recognition Tool in Matlab®. The best performance with the features extracted only from sEMG data was 94.40% using the Artificial Neural Networks (ANN) method. The best performance with the features extracted from both sEMG and gyroscopic data was 96.30% (p-value < 0.05)with the ANN method. It is seen that gyroscopic signals can increase classification performance.

Cell Recognition Using BP Neural Network Edge Computing.

This exploration is to solve the efficiency and accuracy of cell recognition in biological experiments. Neural network technology is applied to the research of cell image recognition. The cell image recognition problem is solved by constructing an image recognition algorithm. First, with an in-depth understanding of computer functions, as a basic intelligent algorithm, the artificial neural network (ANN) is widely used to solve the problem of image recognition. Recently, the backpropagation neural network (BPNN) algorithm has developed into a powerful pattern recognition tool and has been widely used in image edge detection. Then, the structural model of BPNN is introduced in detail. Given the complexity of cell image recognition, an algorithm based on the ANN and BPNN is used to solve this problem. The BPNN algorithm has multiple advantages, such as simple structure, easy hardware implementation, and good learning effect. Next, an image recognition algorithm based on the BPNN is designed and the image recognition process is optimized in combination with edge computing technology to improve the efficiency of algorithm recognition. The experimental results show that compared with the traditional image pattern recognition algorithm, the recognition accuracy of the designed algorithm for cell images is higher than 93.12%, so it has more advantages for processing the cell image algorithm. The results show that the BPNN edge computing can improve the scientific accuracy of cell recognition results, suggesting that edge computing based on the BPNN has a significant practical value for the research and application of cell recognition.

Pattern Recognition for the Electronic Phase of Bismuth Antimony Thin Films

There are many applications involving the use of bismuth antimony thin films. However, due to the low crystalline symmetry and strong coupling between the electronic band edges, it has always been challenging to infer the electronic phase of such a material. Fortunately, with the development of pattern recognition technology, scientists can build many black-box tools for predicting various materials properties. In this present work, we have developed several pattern recognition tools to predict the electronic phase of a bismuth antimony thin film. The support vector machine, the decision tree, and the artificial neural network are used to achieve a prediction accuracy of ~90%, ~95% and ~100%, respectively.

Progress of Machine Learning in the Field of Intrusion Detection Systems

With the growth in the use of the Internet and local area networks, malicious attacks and intrusions into computer systems are increasing. Implementing intrusion detection systems have become extremely important to help maintain good network security. Support vector machines (SVMs), a classic pattern recognition tool, have been widely used in intrusion detection. They can handle very large data with high efficiency, are easy to use, and exhibit good prediction behavior. This paper presents a new SVM model enriched with a Gaussian kernel function based on the features of the training data for intrusion detection. The new model is tested with the CICIDS2017 dataset. The test proves better results in terms of detection efficiency and false alarm rate, which can give better coverage and make detection more efficient.

Editorial on the Citizen Science and Geospatial Capacity Building

This article introduces the Special Issue on “Citizen Science and Geospatial Capacity Building” and briefly evaluates the future trends in this field. This Special Issue was initiated for emphasizing the importance of citizen science (CitSci) and volunteered geographic information (VGI) in various stages of geodata collection, processing, analysis and visualization; and for demonstrating the capabilities and advantages of both approaches. The topic falls well within the main focus areas of ISPRS Commission V on Education and Outreach. The articles collected in the issue have shown the enormously wide application fields of geospatial technologies, and the need of CitSci and VGI support for efficient information extraction and synthesizing. They also pointed out various problems encountered during these processes. The needs and future research directions in this subject can broadly be categorized as; (a) data quality issues especially in the light of big data; (b) ontology studies for geospatial data suited for diverse user backgrounds, data integration, and sharing; (c) development of machine learning and artificial intelligence based online tools for pattern recognition and object identification using existing repositories of CitSci and VGI projects; and (d) open science and open data practices for increasing the efficiency, decreasing the redundancy, and acknowledgement of all stakeholders.

Data-Driven Prognosis of the Failure of Lithium-Ion Batteries

Although the electric vehicle market is witnessing an unprecedented evolution, the fast adoption of these vehicles requires a more thorough status analysis of the battery performance's functionality and reliability. Due to their rechargeable nature, Lithium-ion batteries (LIBs) operation is subject to different irreversible processes during their charging and discharging cycles and causing capacity fade due to various degradation mechanisms. These processes generally result in battery capacity degradation, which usually results in battery failure, with consequences ranging from loss of operation, reduced capability, downtime, and catastrophic malfunctions. To address the issues mentioned above, numerous studies have been dedicated to proposing proper degradation model mechanisms for improving the reliability and availability of LIBs. However, due to accuracy and computational complexity challenges, most existing remaining useful life (RUL) and health prediction models focus on special degradation effects and ignore the integrated deterioration mechanisms, which generally involve batteries' capacity fade associated with the inadequacy of current health estimation tools. Thus, these shortcomings ultimately echo the need to devise novel tools to identify the dominant criteria negatively affecting battery performance and accurately predict the system's failure. The above challenges eventually necessitate a robust and reliable predictive or prognostic capability for prognostics and health monitoring (PHM) under a complexly hostile working environment. In this context, this investigation aims at proposing a novel data-driven approach called data-driven prognosis (DDP) that estimates the relevant constitutive parameters in situ and captures deviations from the expected degradation dynamics of the LIBs in addition to precise modeling of the degradation and capacity models. This talk will present a new data-driven approach using statistical pattern recognition and machine learning tools to detect batteries' anomalies and failures.

Kohonen Artificial Neural Network and Multivariate Analysis in the Identification of Proteome Changes during Early and Long Aging of Bovine Longissimus dorsi Muscle Using SWATH Mass Spectrometry.

To study proteomic changes involved in tenderization of Longissimus dorsi, Charolais heifers and bulls muscles were sampled after early and long aging (12 or 26 days). Sensory evaluation and instrumental tenderness measurement were performed. Proteins were analyzed by gel-free proteomics. By pattern recognition (principal component analysis and Kohonen’s self-organizing maps) and classification (partial least squares-discriminant analysis) tools, 58 and 86 dysregulated proteins were detected after 12 and 26 days of aging, respectively. Tenderness was positively correlated mainly with metabolic enzymes (PYGM, PGAM2, TPI1, PGK1, and PFKM) and negatively with keratins. Downregulation in hemoglobin subunits and carbonic anhydrase 3 levels was relevant after 12 days of aging, while mimecan and collagen chains levels were reduced after 26 days of aging. Bioinformatics indicated that aging involves a prevalence of metabolic pathways after late and long periods. These findings provide a deeper understanding of changes involved in aging of beef and indicate a powerful method for future proteomics studies.

Identification of olfactory characteristics of edible oil during storage period using metal oxide semiconductor sensor signals and ANN methods

In this study, an electronic nose coupled with artificial neural network (ANN) was used to predict the shelf life of two oil with new production date and oil with old production date over a period of 150 days. According to the American Oil Chemists’ Society results, the oils were oxidized after 60 days. Principal component analysis results indicated that all the oil samples were correctly discriminated from each other during their storage times, and samples of oxidized and nonoxidized oils can be properly distinguished from each other. Two main components (PC1, PC2) managed to describe 97% of the data set variance concerning the shelf life of the oil. To develop the ANN models, the data were first divided into three groups: training (60%), validation, and test (40%). To determine the best model, two criteria (R2 and root mean square error) were used. The results revealed that the ANN model can be used as a powerful tool for pattern recognition and determination of the shelf life of oil and its oxidation degree at high precision. Scientific and feasible results can be obtained by matching ANN and the results obtained by metal oxide semiconductor sensors of E-nose. Practical applications One of the most important causes of food spoilage is lipid oxidation. The American Oil Chemists’ Society (AOCS) has developed a variety of methods for assessing the state of food oxidation. In this study, oil shelf life studied by a combination of artificial senses and chemometrics methods. The acidity, peroxide, anisidine, and Totox values of the edible oil samples were measured according to the AOCS standard. Principal component analysis and artificial neural neural methods succeeded in classifying the samples based on their storage time with high accuracy.

Enhancing Thermal Performance and Lifetime Cycles of Li-ion Battery in Electric Vehicles

Hybrid energy storage system has essential priority in Electric Vehicle applications. Therefore, the design of an appropriate power sharing algorithm among energy storage components is necessary to improve battery thermal performance and provide extra extension of battery lifetime cycles. This paper presents an analytical study on the effect of using wavelet decomposition-based power sharing algorithm to force the high frequency component to be fed by the supercapacitor and accordingly reduces the thermal stress on the battery. The proposed approach was investigated by applying it on electric vehicle model in ADVISOR Tool/MATLAB using different driving profiles such as Urban Dynamometer Driving Schedule profile, Highway Fuel Economy Test, New York City Cycle, Los Angeles 1992 and new European driving cycle. The results declare that by using proposed power sharing algorithm, the working temperature of lithium battery decreases significantly while battery lifetime cycles increase, apparently. For urban dynamometer driving schedule, the operating temperature of lithium battery is improved much at maximum decomposition levels reaching to only 25.6 °C compared to 35 °C. In addition, the battery lifetime cycles increased from 2213 to 2585 cycles. Neural Networks pattern recognition tool is also applied to classify the driving cycle to the nearest reference cycles chosen to represent the different driving conditions which help to detect the appropriate wavelet decomposition level, achieving better battery thermal performance and battery lifetime cycles.

Verified the rapid evaluation of the edible safety of wild porcini mushrooms, using deep learning and PLS-DA.

How to quickly identify poisonous mushrooms is a worldwide problem, because poisonous mushrooms and edible mushrooms have very similar appearances. Even some edible mushrooms must be processed further before they can be eaten. In addition, mushrooms from different geographical origins contain different levels of heavy metals. Eating frequent mushrooms with excessive heavy metal content can also cause food poisoning. This information is very important and needs to be informed to consumers in advance. Through the demand for the safety of porcini mushrooms in the Yunnan area we propose a hierarchical identification system based on Fourier-transform near-infrared (FT-NIR) spectroscopy to evaluate the edible safety of porcini species. We found that deep learning is the most effective means to identify the edible safety of porcini, and the recognition accuracy was 100%, by comparing two pattern recognition tools, deep learning and partial least square discriminant analysis (PLS-DA). Although the accuracy of the PLS-DA test set is 96.10%, the poisonous porcini is not allowed to be wrongly judged. In addition, the cadmium (Cd) content of Leccinum rugosiceps in the Midu area exceeded the standard. Deep learning can trace Le. rugosiceps geographic origin with an accuracy of 100%. The overall results show that deep learning methods based on FT-NIR can identify porcini that is at risk of being eaten. This has useful application prospects in food safety. © 2021 Society of Chemical Industry.

Large-Scale Data Mining of Rapid Residue Detection Assay Data From HTML and PDF Documents: Improving Data Access and Visualization for Veterinarians.

Extra-label drug use in food animal medicine is authorized by the US Animal Medicinal Drug Use Clarification Act (AMDUCA), and estimated withdrawal intervals are based on published scientific pharmacokinetic data. Occasionally there is a paucity of scientific data on which to base a withdrawal interval or a large number of animals being treated, driving the need to test for drug residues. Rapid assay commercial farm-side tests are essential for monitoring drug residues in animal products to protect human health. Active ingredients, sensitivity, matrices, and species that have been evaluated for commercial rapid assay tests are typically reported on manufacturers' websites or in PDF documents that are available to consumers but may require a special access request. Additionally, this information is not always correlated with FDA-approved tolerances. Furthermore, parameter changes for these tests can be very challenging to regularly identify, especially those listed on websites or in documents that are not publicly available. Therefore, artificial intelligence plays a critical role in efficiently extracting the data and ensure current information. Extracting tables from PDF and HTML documents has been investigated both by academia and commercial tool builders. Research in text mining of such documents has become a widespread yet challenging arena in implementing natural language programming. However, techniques of extracting tables are still in their infancy and being investigated and improved by researchers. In this study, we developed and evaluated a data-mining method for automatically extracting rapid assay data from electronic documents. Our automatic electronic data extraction method includes a software package module, a developed pattern recognition tool, and a data mining engine. Assay details were provided by several commercial entities that produce these rapid drug residue assay tests. During this study, we developed a real-time conversion system and method for reflowing contents in these files for accessibility practice and research data mining. Embedded information was extracted using an AI technology for text extraction and text mining to convert to structured formats. These data were then made available to veterinarians and producers via an online interface, allowing interactive searching and also presenting the commercial test assay parameters in reference to FDA-approved tolerances.

The use of electronic tongue (e-tongue) as a simple and rapid method for honey authentication

A simple and rapid test for honey authentication is required particularly for the food industry to assure the quality of honey. The conventional methods for honey authentication are costly, requires a long waiting time to obtain results, requires highly skilled personnel, and is difficult in terms of sample preparation. The electronic tongue can be utilized as an alternative technique for honey authentication. The electronic tongue frameworks are depended on an array of sensors with low selectivity while being sensitive to several components in the measured sample. The signals gathered by the sensors are processed through pattern recognition tools to produce prediction models that permit the grouping of the samples and the measurement of a portion of their physicochemical properties. Papers that were published from 2015 to 2020 from several databases such as Google Scholar, ScienceDirect, and Pubmed were collected to obtain abstracts and original articles related to the authentication of honey-based on sugar content as well as geographical and botanical origin. This review highlighted the electronic tongue as a simple and rapid test for honey authentication, several original papers also compare the validity of electronic tongue with the high-performance liquid chromatography methods as the gold standard.

A pattern recognition method on smartphones for planar chromatography and verification on chromatograms of four herbal medicines from citrus fruits

Image retrieval techniques in combination with support vector machine (SVM) on smartphones were studied to make pattern recognition of planar chromatograms. A micellar TLC method employed less toxic and flammable solvents was studied to discriminate four herbal medicines (HMs) from citrus fruits. And the chromatograms were used to test the pattern recognition method. The TLC method employed O/W microemulsion–formic acid (98:2) as the mobile phase, polyamide as the solid phase, and 1% aluminum trichloride aqueous solution as the visualization reagent. TLC of samples of dried lemon peel, dried sweet-orange peel, pummelo peel, and mandarin orange peel of varying origins were performed in three separate runs. Image retrieval techniques combined with SVM directly classified the chromatograms based on migration path images. A mobile application running on an Android smartphone was developed to perform the pattern recognition. The micellar TLC method satisfactorily characterized the chemical profile of the four HMs from citrus fruits. The pattern recognition method combining color layout descriptor with SVM correctly classified all the chromatograms. A smartphone with an application combining color layout descriptor with SVM would be a convenient tool for pattern recognition of planar chromatograms.