Automatic Classification Method Research Articles

MixSleepNet: A Multi-Type Convolution Combined Sleep Stage Classification Model

Background and ObjectiveSleep staging is an essential step for sleep disorder diagnosis, which is time-intensive and laborious for experts to perform this work manually. Automatic sleep stage classification methods not only alleviate experts from these demanding tasks but also enhance the accuracy and efficiency of the classification process. MethodsA novel multi-channel biosignal-based model constructed by the combination of a 3D convolutional operation and a graph convolutional operation is proposed for the automated sleep stages using various physiological signals. Both the 3D convolution and graph convolution can aggregate information from neighboring brain areas, which helps to learn intrinsic connections from the biosignals. Electroencephalogram (EEG), electromyogram (EMG), electrooculogram (EOG) and electrocardiogram (ECG) signals are employed to extract time domain and frequency domain features. Subsequently, these signals are input to the 3D convolutional and graph convolutional branches, respectively. The 3D convolution branch can explore the correlations between multi-channel signals and multi-band waves in each channel in the time series, while the graph convolution branch can explore the connections between each channel and each frequency band. In this work, we have developed the proposed multi-channel convolution combined sleep stage classification model (MixSleepNet) using ISRUC datasets (Subgroup 3 and 50 random samples from Subgroup 1). ResultsBased on the first expert's label, our generated MixSleepNet yielded an accuracy, F1-score and Cohen kappa scores of 0.830, 0.821 and 0.782, respectively for ISRUC-S3. It obtained accuracy, F1-score and Cohen kappa scores of 0.812, 0.786, and 0.756, respectively for the ISRUC-S1 dataset. In accordance with the evaluations conducted by the second expert, the comprehensive accuracies, F1-scores, and Cohen kappa coefficients for the ISRUC-S3 and ISRUC-S1 datasets are determined to be 0.837, 0.820, 0.789, and 0.829, 0.791, 0.775, respectively. ConclusionThe results of the performance metrics by the proposed method are much better than those from all the compared models. Additional experiments were carried out on the ISRUC-S3 sub-dataset to evaluate the contributions of each module towards the classification performance.

3DMASC: Accessible, explainable 3D point clouds classification. Application to bi-spectral topo-bathymetric lidar data

Three-dimensional data have become increasingly present in earth observation over the last decades. However, many 3D surveys are still underexploited due to the lack of accessible and explainable automatic classification methods, for example, new topo-bathymetric lidar data. In this work, we introduce explainable machine learning for 3D data classification using Multiple Attributes, Scales, and Clouds under 3DMASC, a new workflow. This workflow introduces multi-cloud classification through dual-cloud features, encrypting local spectral and geometrical ratios and differences. 3DMASC uses classical multi-scale descriptors adapted to all types of 3D point clouds and new ones based on their spatial variations. In this paper, we present the performances of 3DMASC for multi-class classification of topo-bathymetric lidar data in coastal and fluvial environments. We show how multivariate and embedded feature selection allows the building of optimized predictor sets of reduced complexity, and we identify features particularly relevant for coastal and riverine scene descriptions. Our results show the importance of dual-cloud features, lidar return-based attributes averaged over specific scales, and of statistics of dimensionality-based and spectral features. Additionally, they indicate that small to medium spherical neighbourhood diameters (<7 m) are sufficient to build effective classifiers, namely when combined with distance-to-ground or distance-to-water-surface features. Without using optional RGB information, and with a maximum of 37 descriptors, we obtain classification accuracies between 91 % for complex multi-class tasks and 98 % for lower-level processing using models trained on less than 2000 samples per class. Comparisons with classical point cloud classification methods show that 3DMASC features have a significantly improved descriptive power. Our contributions are made available through a plugin in the CloudCompare software, allowing non-specialist users to create classifiers for any type of 3D data characterized by 1 or 2 point clouds (airborne or terrestrial lidar, structure from motion), and two labelled topo-bathymetric lidar datasets, available on https://opentopography.org/.

Global distribution and dynamics of muddy coasts

Muddy coasts provide ecological habitats, supply food and form a natural coastal defence. Relative sea level rise, changing wave energy and human interventions will increase the pressure on muddy coastal zones. For sustainable coastal management it is key to obtain information on the geomorphology of and historical changes along muddy areas. So far, little is known about the distribution and behaviour of muddy coasts at a global scale. In this study we present a global scale assessment of the occurrence of muddy coasts and rates of coastline change therein. We combine publicly available satellite imagery and coastal geospatial datasets, to train an automated classification method to identify muddy coasts. We find that 14% of the world’s ice-free coastline is muddy, of which 60% is located in the tropics. Furthermore, the majority of the world’s muddy coasts are eroding at rates exceeding 1 m/yr over the last three decades.

Multiscale Hjorth Descriptor on Epileptic EEG Classification

The electroencephalogram (EEG) examination provides information on the brain’s electricity, especially in cases of epilepsy. Since the characteristics of EEG signals are nonlinear and nonstationary, visual inspection becomes very difficult. To overcome this problem, digital EEG signal processing was developed. Automatic epileptic EEG recognition is an area of interest on which much research focuses. The complexity approach to EEG signal analysis is interesting to be used as feature extraction, referring to the nonlinear characteristics of the signal. This study proposed an automatic epileptic EEG classification method based on the multiscale Hjorth descriptor measurement. EEG signals consisting of normal, interictal, and seizure (ictal) were simulated. The signal is scaled into new signals using the coarse-grained procedure on a scale of 1–20. Then, the Hjorth parameter which consists of activity, mobility, and complexity is calculated on the new signal. This process produces a feature vector that is used in the classification stage. Support vector machine (SVM) is used to evaluate the proposed feature extraction method. Simulation results showed that the Hjorth parameter on a scale of 1–15 yields 99.5% accuracy. The proposed method is expected to be applied to digital EEG for seizure detection and prediction.

인공지능 기반의 말더듬 자동분류 방법: 합성곱신경망(CNN) 활용*

인공지능 기반의 말더듬 자동분류 방법: 합성곱신경망(CNN) 활용*

Most soil and litter arthropods are unidentifiable based on current DNA barcode reference libraries.

We are far from knowing all species living on the planet. Understanding biodiversity is demanding and requires time and expertise. Most groups are understudied given problems of identifying and delimiting species. DNA barcoding emerged to overcome some of the difficulties in identifying species. Its limitations derive from incomplete taxonomic knowledge and the lack of comprehensive DNA barcode libraries for so many taxonomic groups. Here, we evaluate how useful barcoding is for identifying arthropods from highly diverse leaf litter communities in the southern Appalachian Mountains (USA). We used 3 reference databases and several automated classification methods on a data set including several arthropod groups. Acari, Araneae, Collembola, Coleoptera, Diptera, and Hymenoptera were well represented, showing different performances across methods and databases. Spiders performed the best, with correct identification rates to species and genus levels of ~50% across databases. Springtails performed poorly, no barcodes were identified to species or genus. Other groups showed poor to mediocre performance, from around 3% (mites) to 20% (beetles) correctly identified barcodes to species, but also with some false identifications. In general, BOLD-based identification offered the best identification results but, in all cases except spiders, performance is poor, with less than a fifth of specimens correctly identified to genus or species. Our results indicate that the soil arthropod fauna is still insufficiently documented, with many species unrepresented in DNA barcode libraries. More effort toward integrative taxonomic characterization is needed to complete our reference libraries before we can rely on DNA barcoding as a universally applicable identification method.

Creating a software application to help university educators to reflect on the cognitive complexity of their exam questions, using Bloom’s Taxonomy and automated classification

Previous research has shown that many university educators struggle to accurately evaluate the cognitive complexity of exam questions (and overall exams) which they write, based on Bloom's Taxonomy. This can lead to concerns about the design of exams. Software tools could possibly assist educators via automated classification methods. This paper reports a work-in-progress project that is creating a software application (tool) to assist university educators with writing exams. We evaluate 3 methods of automated classification including keywords-based approach, OpenAI evaluation, and an existing algorithm. The tool is designed to be able to help educators to reflect on their exam, by providing educators with meaningful feedback on question complexity and the overall exam, assisting in exam design. The software tool developed in this study is expected to benefit educators by providing objective feedback and serving as a professional development resource.

Evaluating sleep-stage classification: how age and early-late sleep affects classification performance.

Sleep stage classification is a common method used by experts to monitor the quantity and quality of sleep in humans, but it is a time-consuming and labour-intensive task with high inter- and intra-observer variability. Using wavelets for feature extraction and random forest for classification, an automatic sleep-stage classification method was sought and assessed. The age of the subjects, as well as the moment of sleep (early-night and late-night), were confronted to the performance of the classifier. From this study, we observed that these variables do affect the automatic model performance, improving the classification of some sleep stages and worsening others.

Eddy Currents Assessment of Rail Cracks Using Artificial Neural Networks in a Laboratory Setup

Although flaws associated with rolling contact fatigue (RCF) and the corresponding traffic induced damage, which are a cause of failure in railways, have been of great concern in railway system maintenance and safety strategies in many countries for at least two decades, this serious problem has not been yet adequately tackled in the Argentine railway system. The present upgrading activities undertaken in the Argentine railway system (in infrastructure and in rolling stock) are prompting the need for R&amp;D in non-destructive testing techniques and procedures, to satisfy requirements of the new rolling stock and to ensure safe and economic operation of passengers and cargo. RCF damage appears as surface and near surface defects and grows into cracks which in time will propagate along the running surface and through the cross-section. Eddy current testing (ET) is a very efficient in-service inspection method for this task, the near field technique being especially recommended for ferromagnetic components.&#x0D; In the present paper, an artificial neural network (ANN) method for automatic classification of flaws with lift-off compensation is presented and tested. The tests consist on the ET evaluation of right angle artificial cracks on a rail calibration coupon; the depth of the cracks studied ranged from 1 to 7 mm. The technique permitted to compensate the weakening of the signals caused by the lift-off effect, allowing signal cracks classification with lift-off variations of up to 5.4 mm.&#x0D; The effect of crack skewness on the ET signals is also studied. Because the RCF cracks penetrate the rail at oblique angles, (10° to 30° to the rolling surface), an additional uncertainty component is added to the experiments if calibration is made with a piece having perpendicular cracks. In order to estimate this additional uncertainty on the ANN method presented here, further tests were made with a second calibration piece with cracks at 25° to the surface. Comparison of results showed that the peak to peak amplitudes for both types of cracks are not equivalent at all the tested depths.

An automatic defect classification and segmentation method on three-dimensional point clouds for sewer pipes

With the development of deep learning (DL), sewer pipe inspection on two-dimensional (2D) images has achieved remarkable accuracy. However, extracting defect measurements from these 2D images is challenging due to the curved nature of pipes and the lack of depth information. Point clouds can restore the three-dimensional (3D) information of objects. To effectively identify defects in disordered and sparse point clouds, a 3D sewer pipe classification and segmentation method was proposed. In the encoder, the original point clouds are sampled and grouped and the local features in the clusters are extracted by two symmetric functions (1 × 1 convolution and the maximization function) to process the points with permutation invariance. In the decoder, the multi-scaling abstract features are upsampled using feature pyramid network (FPN) to predict the category of each point. Especially, the network structure and training strategy of the inspection method is optimized to improve the inspection accuracy. Furthermore, two data augmentation methods, namely random scaling and point jitter, are used to increase the data volume. An ablation experiment shows that the optimization of network structure can effectively improve the performance of the inspection model and the novel training strategies can stabilize the training process and prevent overfitting. Comparison among the state-of-the-art networks demonstrates that the proposed segmentation model attains the highest mIoU of 94.15 %, which is improved by 11.46 % with the optimization of network structure and training strategy. For the classification task, the F1 score and accuracy of the established model are 6.79 % and 5.46 % higher than PointNet++, respectively. These results signify the high-accuracy defect inspection capability of our proposed method on 3D point clouds of sewer pipelines.

Classification of intracranial tumors based on optical-spectral analysis

The motivation for the present study was the need to develop methods of urgent intraoperative biopsy during surgery for removal of intracranial tumors. Based on the experience of previous joint work of GPI RAS and N.N. Burdenko National Medical Research Center of Neurosurgery to introduce fluorescence spectroscopy methods into clinical practice, an approach combining various optical-spectral techniques, such as autofluorescence spectroscopy, fluorescence of 5-ALA induced protoporphyrin IX, diffuse reflection of broadband light, which can be used to determine hemoglobin concentration in tissues and their optical density, Raman spectroscopy, which is a spectroscopic method that allows detection of various molecules in tissues by vibrations of individual characteristic molecular bonds. Such a variety of optical and spectral characteristics makes it difficult for the surgeon to analyze them directly during surgery, as it is usually realized in the case of fluorescence methods – tumor tissue can be distinguished from normal with a certain degree of certainty by fluorescence intensity exceeding a threshold value. In case the number of parameters exceeds a couple of dozens, it is necessary to use machine learning algorithms to build a intraoperative decision support system for the surgeon. This paper presents research in this direction. Our earlier statistical analysis of the optical-spectral features allowed identifying statistically significant spectral ranges for analysis of diagnostically important tissue components. Studies of dimensionality reduction techniques of the optical-spectral feature vector and methods of clustering of the studied samples also allowed us to approach the implementation of the automatic classification method. Importantly, the classification task can be used in two applications – to differentiate between different tumors and to differentiate between different parts of the same (center, perifocal zone, normal) tumor. This paper presents the results of our research in the first direction. We investigated the combination of several methods and showed the possibility of differentiating glial and meningeal tumors based on the proposed optical-spectral analysis method.

Comparison of Automatic Classification Methods for Identification of Ice Surfaces from Unmanned-Aerial-Vehicle-Borne RGB Imagery

This article describes a comparison of the pixel-based classification methods used to distinguish ice from other land cover types. The article focuses on processing RGB imagery, as these are very easy to obtained. The imagery was taken using UAVs and has a very high spatial resolution. Classical classification methods (ISODATA and Maximum Likelihood) and more modern approaches (support vector machines, random forests, deep learning) have been compared for image data classifications. Input datasets were created from two distinct areas: The Pond Skříň and the Baroch Nature Reserve. The images were classified into two classes: ice and all other land cover types. The accuracy of each classification was verified using a Cohen’s Kappa coefficient, with reference values obtained via manual surface identification. Deep learning and Maximum Likelihood were the best classifiers, with a classification accuracy of over 92% in the first area of interest. On average, the support vector machine was the best classifier for both areas of interest. A comparison of the selected methods, which were applied to highly detailed RGB images obtained with UAVs, demonstrates the potential of their utilization compared to imagery obtained using satellites or aerial technologies for remote sensing.

CatBoost-Based Automatic Classification Study of River Network

Existing research on automatic river network classification methods has difficulty scientifically quantifying and determining feature threshold settings and evaluating weights when calculating multi-indicator features of the local and overall structures of river reaches. In order to further improve the accuracy of river network classification and evaluate the feature weight, this paper proposes an automatic grading method for river networks based on ensemble learning in CatBoost. First, the graded river network based on expert knowledge is taken as the case; with the support of the existing case results, a total of eight features from the semantic, geometric, and topological aspects of the river network were selected for calculation. Second, the classification model, obtained through learning and training, was used to calculate the classification results of the main stream and tributaries of the river reach to be classified. Furthermore, the main stream river reaches were connected, and the main stream rivers at different levels were hierarchized to achieve river network classification. Finally, the Shapley Additive explanation (SHAP) framework for interpreting machine learning models was introduced to test the influence of feature terms on the classification results from the global and local aspects, so as to improve the interpretability and transparency of the model. Performance evaluation can determine the advantages and disadvantages of the classifier, improve the classification effect and practicability of the classifier, and improve the accuracy and reliability of river network classification. The experiment demonstrates that the proposed method achieves expert-level imitation and has higher accuracy for identifying the main stream and tributaries of river networks. Compared with other classification algorithms, the accuracy was improved by 0.85–5.94%, the precision was improved by 1.82–9.84%, and the F1_Score was improved by 0.8–5.74%. In this paper, CatBoost is used for river network classification for the first time, and SHAP is used to explain the influence of characteristics, which improves the accuracy of river network classification and enhances the interpretability of the classification method. By constructing a reasonable hierarchy, a better grading effect can be achieved, and the intelligence level of automatic grading of river networks can be further improved.

Face Mask Image Classification Using Fine-Tuning and The Effect of FGSM And PGD Attacks.

This research seeks to use a fine-tuning method for automatic diagnostics and classification of face masks without face mask images. We trained Deep Convolutional Neural Network models on the face mask dataset obtained from the Mendeley data repository to determine whether individuals will adhere to policies that face mask reduces the spread of SARS-COV-2. The proposed architectures used in this study include the VGG19, MobileNetV3, and InceptionV3 models. These models are known for their role in classifying images. They were trained using the fine-tuning approach and their respective outputs were compared. After training the face mask dataset, it can be said that the fine-tuned InceptionV3 model performed extremely well against the other models obtaining an accuracy of 99.21%, and was able to predict 99.63% of the test dataset. However, the robustness of the fine-tuned model was tested against fast gradient sign method (FGSM) and projected gradient descent (PGD) attacks which generate adversarial images using the gradients of the model. Additionally, the classification report shows after the FGSM attack that the model's accuracy was reduced by 43%. Also, after the PDG attack, the accuracy of the model was reduced by 19%. We assessed the models using many performance metrics such as precision, recall, F1 score, and accuracy before subjecting them to two common adversarial attack techniques: FGSM and PDG. Finally, we demonstrated how the proposed robust method improved the model’s defense against adversarial attacks. The findings emphasize the critical need to increase awareness about adversarial assaults on SARS-COV-2 monitoring systems and to advocate for proactive steps to protect healthcare systems from comparable threats prior to practical deployment.

Support Vector Machine based Automatic Classification Method for IoT big Data Features

&lt;p&gt;As China’s information technology development shifts from a single high-speed growth stage to a multidimensional high-quality development stage, the Internet of Things (IoT) enters all aspects of life and becomes more and more popular. The demand for IoT big data information analysis and processing is increasing, and the important role of feature automatic classification methods becomes increasingly prominent. This research proposes SPO-SVM and WSPO-SVM models based on support vector machine for smart home environment monitoring data under the big data of Internet of Things, and then optimizes them with particle swarm optimization algorithm and adaptive method. Finally, the data set is selected for comparative experimental analysis of each optimization algorithm model. The experimental results show that the optimized WSPO-SVM model has less total misclassification and single class misclassification compared with other algorithms under Wine dataset. In cross-validation, both its classification accuracy performance outperformed other algorithms. Under 10 sets of smart home environment monitoring data sets, the WSPO-SVM algorithm model achieves 100% accuracy in 6 out of 10 test data sets, with an average accuracy of 97.67%, which is about 9% higher than the ordinary SVM algorithm model and about 15% higher than other feature classification algorithms. The experimental results prove that the WSPO-SVM algorithm can complete the feature classification work in the IoT big data environment, which meets the expectation.&lt;/p&gt; &lt;p&gt;&amp;nbsp;&lt;/p&gt;

Classification of the tumor immune microenvironment and associations with outcomes in patients with metastatic melanoma treated with immunotherapies

BackgroundTumor microenvironment (TME) characteristics are potential biomarkers of response to immune checkpoint inhibitors in metastatic melanoma. This study developed a method to perform unsupervised classification of TME of metastatic melanoma.MethodsWe...

An efficient Automatic Modulation Classification method based on the Convolution Adaptive Noise Reduction network

Due to the influence of noise in the received signal in non-cooperative communication, it is difficult for existing Automatic modulation classification methods to balance classification accuracy and model complexity. This paper proposes a novel Convolutional Adaptive Noise Reduction (CANR) network, which consists of an Adaptive Noise Reduction (ANR) module and a Convolutional Feature Extraction (CFE) module. The ANR and CFE modules denoise the combined input and capture the spatiotemporal features in the time series. Experiments on benchmark datasets show that the proposed network has the fewest training parameters and state-of-the-art recognition accuracy under the same conditions.

The potential of automated classification to categorise police force tweeting behaviours: leading the way to large scale analysis

ABSTRACT Police activity on social media has emerged as a significant and expanding area of research. However, the existing body of research has predominantly adopted qualitative methods or focused on small-scale samples for quantitative analysis. This study presents a novel approach to analysing police social media behaviours, employing automated classification methods to generate a substantial sample of categorised police tweets. Encompassing over 40,000 tweets from five United Kingdom forces, collected over a three-year period, this dataset represents one of the largest evaluated samples in the domain of police social media research. A core objective of this research is to investigate the extent to which police tweeting behaviours align with three common categories identified in the literature: providing information, engagement, and intelligence gathering. To achieve this, a two-pronged methodology is employed, combining manual content analysis and an applied automated classification approach. This comprehensive method aims to create a sample of classified police tweets, effectively representing their diverse tweeting behaviours. The classicisation process involves the training and testing of three automated models, namely naïve Bayes, logistic regression, and XGBoost, evaluating the accuracy of their results to ensure a robust and reliable classification outcome. Furthermore, the resulting sample is subject to additional in-depth analyses. The exploration encompasses various facets of tweet content, style, overall usage, and adaptability across different police forces. Additionally, the research considers public interactions with the police tweets. These analyses are conducted for each force and class, thereby establishing connections between social media interactions and their potential impact on highlighted agendas.

Automatic Detection of Plateau Phases in Light Curves of Variable Stars

Modern astronomical surveys produce millions of light curves of variable sources. These massive data sets challenge the community to create automatic light-curve processing methods for detection, classification, and characterization of variable stars. In this paper, we present a novel method for extracting the variable components of a light curve based on Otsu’s thresholding method. To validate the effectiveness of this method, we apply it to the light curves of detached eclipsing binaries and dwarf novae, sourced from Optical Gravitational Lensing Experiment catalogs.

MobileRaT: A Lightweight Radio Transformer Method for Automatic Modulation Classification in Drone Communication Systems

Nowadays, automatic modulation classification (AMC) has become a key component of next-generation drone communication systems, which are crucial for improving communication efficiency in non-cooperative environments. The contradiction between the accuracy and efficiency of current methods hinders the practical application of AMC in drone communication systems. In this paper, we propose a real-time AMC method based on the lightweight mobile radio transformer (MobileRaT). The constructed radio transformer is trained iteratively, accompanied by pruning redundant weights based on information entropy, so it can learn robust modulation knowledge from multimodal signal representations for the AMC task. To the best of our knowledge, this is the first attempt in which the pruning technique and a lightweight transformer model are integrated and applied to processing temporal signals, ensuring AMC accuracy while also improving its inference efficiency. Finally, the experimental results—by comparing MobileRaT with a series of state-of-the-art methods based on two public datasets—have verified its superiority. Two models, MobileRaT-A and MobileRaT-B, were used to process RadioML 2018.01A and RadioML 2016.10A to achieve average AMC accuracies of 65.9% and 62.3% and the highest AMC accuracies of 98.4% and 99.2% at +18 dB and +14 dB, respectively. Ablation studies were conducted to demonstrate the robustness of MobileRaT to hyper-parameters and signal representations. All the experimental results indicate the adaptability of MobileRaT to communication conditions and that MobileRaT can be deployed on the receivers of drones to achieve air-to-air and air-to-ground cognitive communication in less demanding communication scenarios.