Linear Support Vector Machine Research Articles

Enhancing Depression Detection: A Stacked Ensemble Model with Feature Selection and RF Feature Importance Analysis Using NHANES Data

Around the world, 5% of adults suffer from depression, which is often inadequately treated. Depression is caused by a complex relationship of cultural, psychological, and physical factors. This growing issue has become a significant public health problem globally. Medical datasets often contain redundant characteristics, missing information, and high dimensionality. By using an iterative floating elimination feature selection algorithm and considering various factors, we can reduce the feature set and achieve optimized outcomes. The research utilizes the 36-Item Short Form Survey (SF-36) from the NHANES 2015–16 dataset, which categorizes data into seven groups relevant to quality of life and depression. This dataset presents a challenge due to its imbalance, with only 8.08% of individuals diagnosed with depression. The Depression Ensemble Stacking Generalization Model (DESGM) employs stratified k-fold cross-validation and oversampling for training data. DESGM enhances the classification performance of both base learners (linear support vector machine, perceptron, artificial neural network, linear discriminant analysis, and K-nearest neighbor) and meta-learners (logistic regression). The model achieved an F1 score of 0.9904 and an accuracy of 98.17%, with no instances of depression misdiagnosed.

Enhancing early identification of high-fertile cattle females using infrared blood serum spectra and machine learning

Artificial insemination (AI) success in bovine reproduction is vital for the cattle industry’s economic sustainability and for advancing the understanding of reproductive physiology. Identify high-fertile animals’ fertility is a complex task due to multifactorial traits, including hormonal, age-related, and body condition factors. Early high-fertility identification is crucial for timely interventions and enhancing AI success. In this study, we present the potential use of Fourier-transform infrared (FTIR) spectroscopy on blood serum for early identification of high-fertile Nellore female cows for AI protocols. Blood serum FTIR spectra were obtained from Nellore female cows before AI. FTIR spectra underwent data analysis and the results demonstrated successful discrimination between animals that exhibit pregnant and non-pregnant diagnoses 30 days after AI. FTIR spectra revealed consistent vibrational modes, emphasizing Amide I and II bands. Principal Component Analysis (PCA) effectively segregated groups based on molecular information. Linear SVM with C = 10 and 4 PCs achieved 100% accuracy in the group classification. This innovative approach using FTIR spectroscopy and ML algorithms offers a promising means of high-fertile cow identification, potentially improving AI outcomes in Nellore cattle. The study presents valuable insights into advancements in reproductive management practices for this economically significant breed.

Admission blood tests predicting survival of SARS-CoV-2 infected patients: a practical implementation of graph convolution network in imbalance dataset

BackgroundPredicting an individual’s risk of death from COVID-19 is essential for planning and optimising resources. However, since the real-world mortality rate is relatively low, particularly in places like Hong Kong, this makes building an accurate prediction model difficult due to the imbalanced nature of the dataset. This study introduces an innovative application of graph convolutional networks (GCNs) to predict COVID-19 patient survival using a highly imbalanced dataset. Unlike traditional models, GCNs leverage structural relationships within the data, enhancing predictive accuracy and robustness. By integrating demographic and laboratory data into a GCN framework, our approach addresses class imbalance and demonstrates significant improvements in prediction accuracy.MethodsThe cohort included all consecutive positive COVID-19 patients fulfilling study criteria admitted to 42 public hospitals in Hong Kong between January 23 and December 31, 2020 (n = 7,606). We proposed the population-based graph convolutional neural network (GCN) model which took blood test results, age and sex as inputs to predict the survival outcomes. Furthermore, we compared our proposed model to the Cox Proportional Hazard (CPH) model, conventional machine learning models, and oversampling machine learning models. Additionally, a subgroup analysis was performed on the test set in order to acquire a deeper understanding of the relationship between each patient node and its neighbours, revealing possible underlying causes of the inaccurate predictions.ResultsThe GCN model was the top-performing model, with an AUC of 0.944, considerably outperforming all other models (p < 0.05), including the oversampled CPH model (0.708), linear regression (0.877), Linear Discriminant Analysis (0.860), K-nearest neighbours (0.834), Gaussian predictor (0.745) and support vector machine (0.847). With Kaplan-Meier estimates, the GCN model demonstrated good discriminability between low- and high-risk individuals (p < 0.0001). Based on subanalysis using the weighted-in score, although the GCN model was able to discriminate well between different predicted groups, the separation was inadequate between false negative (FN) and true negative (TN) groups.ConclusionThe GCN model considerably outperformed all other machine learning methods and baseline CPH models. Thus, when applied to this imbalanced COVID survival dataset, adopting a population graph representation may be an approach to achieving good prediction.

Application of computer vision and machine learning in morphological characterization of Adansonia digitata fruits

Measuring fruit mass and volume is a time-consuming and tedious task that can affect production planning. This study developed a computer vision system to estimate the volume and mass of baobab fruits from single-view images captured from inexpensive and readily available cameras such as those in smartphones. The baobab fruits were collected from two study fields within the savanna ecological zone. Their images were captured, and subsequently, they were detected and segmented with over 97 % accuracy. The segmented images were binarized, and two-dimensional (2D) features such as the segmented area, centroid, bounding box, equivalent diameter, and major diameter were extracted from them. The volumes of the fruits were estimated from the 2D features using random forest, linear, polynomial, and radial support vector machine models. All the models achieved high goodness of fit; however, the random forest model delivered the best performance, with an R2 value of 99.8 %. The relationship between mass and volume was a quadratic equation (mass = 38.23 + 0.25 × volume + 4.49e−05 × volume2) and had an R2 value of 92 %. Correlations were validated via plots and statistical tests, and credible intervals of point estimates were determined from the posterior distributions of their samples. This highlights the potential of artificial intelligence methods to be applied in a less constrained environmental setting for ecological research and agricultural management. Commercial companies producing baobab powder and seed oil should apply these models for effective production planning. To enhance the model, it would be beneficial to gain a better understanding of how climate gradients affect the morphological characteristics of baobab fruits.

A novel voice classification based on Gower distance for Parkinson disease detection

BackgroundTraditional classifier for the classification of diseases, such as K-Nearest Neighbors (KNN), Linear Discriminant Analysis (LDA), Random Forest (RF), Logistic Regression (LR), and Support Vector Machine (SVM), often struggle with high-dimensional medical datasets. ObjectiveThis study presents a novel classifier to overcome the limitations of traditional classifiers in Parkinson’s disease (PD) detection based on Gower distance. MethodsWe present the Gower distance metric to handle diverse feature sets in voice recordings, which acts as a dissimilarity measure for all feature types, making the model adept at identifying subtle patterns indicative of PD. Additionally, the Cuckoo Search algorithm is employed for feature selection, reducing dimensionality by focusing on key features, thereby lessening the computational load associated with high-dimensional datasets. ResultsThe proposed classifier based on Gower distance resulted in an accuracy rate of 98.3% with feature selection and achieved an accuracy of 94.92% without the feature selection method. It outperforms traditional classifiers and recent studies in PD detection from voice recordings. ConclusionsThis accuracy shows the capability of the approach in the correct classification of instances and points out the potential of the approach as a reliable diagnostic tool for the medical practitioner. The findings state that the proposed approach holds promise for improving the diagnosis and monitoring of PD, both within medical institutions and at homes for the elderly.

Folk art classification using support vector machine

&lt;p dir="ltr"&gt;Tremendous amounts of effort have been carried out every year by the governments of all the countries to preserve art and culture. Art in the form of paintings, artifacts, music, dance, and cuisines of every country has the utmost importance. The study of Tribal arts provides deep insight into our history and acts as a milestone in the roadmap of our future. This paper focuses on three popular folk arts namely: Gond, Manjusha, and Warli. 300 images of each artwork have been collected from various online repositories. To generate a robust system, data augmentation is applied which results in 7510 images. A feature vector based on a generalized co-occurrence matrix, local binary pattern, HSV histogram, and canny edge detector is constructed and classification is performed using a linear support vector machine. 10- fold cross-validation produces 99.8% accuracy.&lt;/p&gt;

Machine learning for real estate valuation: Astana, Kazakhstan case

Purpose of this research is to investigate the accuracy of machine learning models in forecasting and evaluating house prices, and to understand the key factors that impact pricing. The study involved analyzing data scraped from real estate ads in the “sale of secondary housing” category on the website krisha.kz. The paper emphasizes the importance of understanding the factors that affect house prices, such as quality, location, size, and building materials. It was concluded that these factors have a strong correlation with house price prediction. The information available on krisha.kz was found to be a useful resource for finding good apartments. The data collected by the scraper was analyzed by models: Linear regression (LR), interactions linear regression (ILR), robust linear regression (RLR), fine tree regression (FTR), medium tree regression (MTR), coarse tree regression (CTR), linear support vector machine (LSVM), quadratic SVM (QSVM), medium gaussian SVM (MGSVM), rational quadratic gaussian process regression (RQGPR), boosted trees (BoosT), bagged trees (BagT), neural network based on the bayesian regularization algorithm (BR-BPNN). BR-BPNN showed better results than other models, with an MSE of 32.14 and R of 0.9899.

Performance enhancement of classifiers through Bio inspired feature selection methods for early detection of lung cancer from microarray genes

Gene expression in the microarray is assimilated with redundant and high-dimensional information. Moreover, the information in the microarray genes mostly correlates with background noise. This paper uses dimensionality reduction and feature selection methods to employ a classification methodology for high-dimensional lung cancer microarray data. The approach is enforced in two phases; initially, the genes are dimensionally reduced through Hilbert Transform, Detrend Fluctuation Analysis and Least Square Linear Regression methods. The dimensionally reduced data is further optimized in the next phase using Elephant Herd optimization (EHO) and Cuckoo Search Feature selection methods. The classifiers used here are Bayesian Linear Discriminant, Naive Bayes, Random Forest, Decision Tree, SVM (Linear), SVM (Polynomial), and SVM (RBF). The classifier's performances are analysed with and without feature selection methods. The SVM (Linear) classifier with the DFA Dimensionality Reduction method and EHO feature selection achieved the highest accuracy of 92.26 % compared to other classifiers.

Electroencephalography microstate alterations reflect potential double-edged cognitive adaptation in Ménière's disease.

To explore the microstate characteristics and underlying brain network activity of Ménière's disease (MD) patients based on high-density electroencephalography (EEG), elucidate the association between microstate dynamics and clinical manifestation, and explore the potential of EEG microstate features as future neurobiomarkers for MD. Thirty-two patients diagnosed with MD and 29 healthy controls (HC) matched for demographic characteristics were included in the study. Dysfunction and subjective symptom severity were assessed by neuropsychological questionnaires, pure tone audiometry, and vestibular function tests. Resting-state EEG recordings were obtained using a 256-channel EEG system, and the electric field topographies were clustered into four dominant microstate classes (A, B, C, and D). The dynamic parameters of each microstate were analyzed and utilized as input for a support vector machine (SVM) classifier to identify significant microstate signatures associated with MD. The clinical significance was further explored through Spearman correlation analysis. MD patients exhibited an increased presence of microstate class C and a decreased frequency of transitions between microstate class A and B, as well as between class A and D. The transitions from microstate class A to C were also elevated. Further analysis revealed a positive correlation between equilibrium scores and the transitions from microstate class A to C under somatosensory challenging conditions. Conversely, transitions between class A and B were negatively correlated with vertigo symptoms. No significant correlations were detected between these characteristics and auditory test results or emotional scores. Utilizing the microstate features identified via sequential backward selection, the linear SVM classifier achieved a sensitivity of 86.21% and a specificity of 90.61% in distinguishing MD patients from HC. We identified several EEG microstate characteristics in MD patients that facilitate postural control yet exacerbate subjective symptoms, and effectively discriminate MD from HC. The microstate features may offer a new approach for optimizing cognitive compensation strategies and exploring potential neurobiological markers in MD.

Wavelet-Based precipitation preprocessing for improved drought Forecasting: A Machine learning approach using tunable Q-factor wavelet transform and maximal overlap discrete wavelet transform

Drought forecasting plays a crucial role in mitigating the severe agricultural and social consequences caused by droughts. The fluctuating nature of droughts makes it difficult to develop an effective drought forecasting model without preprocessing the input data. This paper proposes a novel approach that introduces the tunable Q-factor wavelet transform (TQWT) with the maximal overlap discrete wavelet transform (MODWT) based Fejér–Korovkin, Coiflet, and Daubechies filters in the decomposition of precipitation data for the extended lead time forecasting of the standardized precipitation evapotranspiration index (SPEI). The decomposed datasets have been coupled with Matern Gaussian process regression (MGPR), exponential Gaussian process regression (EGPR), linear support vector machine (LSVM), and coarse Gaussian support vector machine (CGSVM), and formed hybrid models to forecast SPEI-12 and SPEI-18 for several lead times (i.e., 6, 12, 18, and 24 months). Results of the study represent that the wavelet-based hybrid models are capable of predicting SPEI-12 and SPEI-18 effectively for different lead times with promising results. Both TQWT and MODWT coupled with MGPR yielded reasonable performances for the lead time of 6 months in all stations. However, for the higher lead times, TQWT coupled with MGPR outperformed other hybrid models. The results of the TQWT-MGPR for SPEI-12 are more effective than SPEI-18 in different lead times. The study highlights that preprocessing of precipitation data using TQWT is a promising direction for drought forecasting, and the findings obtained from drought forecasting can be utilized in the areas of water and agricultural resource management to effectively mitigate and alleviate the potential impacts of future droughts.

Altered Resting-State Brain Entropy in Cerebral Small Vessel Disease Patients with Cognitive Impairment.

Objective: Cerebral small vessel disease (CSVD) is a primary vascular disease of cognitive impairment. Previous studies have predominantly focused on brain linear features. However, the nonlinear measure, brain entropy (BEN), has not been elaborated. Thus, this study aims to investigate if BEN abnormalities could manifest in CSVD patients with cognitive impairment. Methods: Thirty-four CSVD patients with cognitive impairment and 37 healthy controls (HCs) were recruited. Analysis of gray matter approximate entropy (ApEn) and sample entropy (SampEn) which are two indices of BEN was calculated. To explore whether BEN can provide unique information, we further performed brain linear methods, namely, amplitude of low frequency fluctuation (ALFF) and regional homogeneity (ReHo), to observe their differences. The ratios of BEN/ALFF and BEN/ReHo which represent the coupling of nonlinear and linear features were introduced. Correlation analysis was conducted between imaging indices and cognition. Subsequently, the linear support vector machine (SVM) was used to assess their discriminative ability. Results: CSVD patients exhibited lower ApEn and SamEn values in sensorimotor areas, which were correlated with worse memory and executive function. In addition, the results of BEN showed little overlap with ALFF and ReHo in brain regions. Correlation analysis also revealed a relationship between the two ratios and cognition. SVM analysis using BEN and its ratios as features achieved an accuracy of 74.64% (sensitivity: 86.49%, specificity: 61.76%, and AUC: 0.82439). Conclusion: Our study reveals that the reduction of sensorimotor system complexity is correlated with cognition. BEN exhibits distinctive characteristics in brain activity. Combining BEN and the ratios can be new biomarkers to diagnose CSVD with cognitive impairment. Impact Statement Cerebral small vessel disease (CSVD) is regarded as the most important vascular disease of cognitive impairment. However, conventional brain imaging fails to adequately elucidate the pathogenesis of cognitive disorder related to CSVD. In this regard, exploring brain entropy (BEN) based on resting-state functional magnetic resonance imaging (rs-fMRI) represents a relatively novel and unexplored approach in the context of CSVD. This approach provides novel insights into the pathogenesis, diagnosis, and rehabilitation of cognitive disorder associated with CSVD.

Unveiling individuality in the early phase of motor learning: a machine learning approach for analysing weightlifting technique in novices.

Despite the growing body of evidence highlighting the individuality in movement techniques, predominant models of motor learning, particularly during the acquisition phase, continue to emphasise generalised, person-independent approaches. Biomechanical studies, coupled with machine learning approaches, have demonstrated the uniqueness of movement techniques exhibited by individuals. However, this evidence predominantly pertains to already stabilised movement techniques, particularly evident in cyclic daily activities such as walking, running, or cycling, as well as in expert-level sports movements. This study aims to evaluate the hypothesis of individuality in whole-body movements necessitating intricate coordination and strength among novice participants at the very beginning of an acquisition phase. In a within-subject design, sixteen highly active male participants (mean age: 23.1 ± 2.1 years), all absolute novices in the learning task (i.e., power snatch of Olympic weightlifting), participated in randomised snatch learning bouts. These bouts comprised 36 trials across various motor learning models: differential learning contextual interference (serial, sCIL; and blocked, bCIL), and repetitive learning. Kinematic and kinetic data were collected from three standardised snatch trials performed following each motor learning model bout. The time-continuous data were input to a linear Support Vector Machine (SVM). We conducted analyses on two classification tasks: participant and motor learning model. The Support Vector Machine classification revealed a notably superior participant classification compared to the motor learning model classification, with an averaged prediction accuracy of 78% (in average ≈35 out of 45 test trials across the folds) versus 27.3% (in average ≈9 out of 36 test trials across the folds). In specific fold and input combinations, accuracies of 91% versus 38% were respectively achieved. Methodically, the crucial role of selecting appropriate data pre-processing methods and identifying the optimal combinations of SVM data inputs is discussed in the context of future research. Our findings provide initial support for a dominance of individuality over motor learning models in movement techniques during the early phase of acquisition in Olympic weightlifting power snatch.

Revealing the Impact of the Combination of Parameters on SVM Performance in COVID-19 Classification

Non-linear SVM functions to modify the kernel in the SVM. Each kernel function in linear and non-linear SVMs has several parameters that are used in the classification process. SVM is a method that has advantages in classification, but there are still obstacles in selecting optimal parameters. This research investigates the effect of parameter variations on SVM classification performance on the COVID-19 dataset, using linear, RBF, Sigmoid and polynomial kernels. The analysis shows that the polynomial kernel is superior with the highest performance compared to other kernels. The highest accuracy of 77.57% was achieved with a combination of C values ??of 0.75 and Gamma of 0.75, and an F1-Score value of 76.67% indicating an optimal balance between precision and recall. The performance stability produced by the polynomial kernel provides advantages in classifying the COVID-19 dataset, with more controlled fluctuations compared to other kernels. The interaction between the C and Gamma parameters shows that a Gamma value of 0.75 consistently provides good results, while adjusting the C parameter shows more controlled performance variations. This confirms that appropriate Gamma parameter settings are key in improving the accuracy and consistency of SVM model predictions in this case.

Face forgery video detection based on expression key sequences

In order to minimize additional computational costs in detecting forged videos, and enhance detection accuracy, this paper employs dynamic facial expression sequences as key sequences, replacing original video sequences as inputs for the detection model. A spatio-temporal dual-branch detection network is designed based on the visual Transformer architecture. Specifically, this process involves three steps. Firstly, dynamic facial expression sequences are localized as key sequences using optical flow difference algorithms. Subsequently, the spatial branch network employs the focal self-attention mechanism to focus on dynamic features of expression-relevant regions and uses Factorization Machines to facilitate feature interaction among multiple key sequences. Meanwhile, the temporal branch network concentrates on learning the temporal inconsistency of optical flow differences between adjacent frames. Finally, a binary classification linear SVM combines the Softmax values from the two branch networks to provide the ultimate detection outcome. Experimental results on the Faceforensics++ dataset demonstrate: (a) replacing whole video sequences with facial expression key sequences effectively reduces training and detection time by nearly 80% and 90%, respectively; (b) compared to state-of-the-art methods involving random sequence/frame extraction and key frame extraction based on video compression techniques, the proposed approach in this paper presents a more competitive detection accuracy.

Classification of LAMOST spectra of B-type and hot subdwarf stars using kernel support vector machine

Machine learning has emerged as a leading field in artificial intelligence, demonstrating expert-level performance in various domains. Astronomy has benefited from machine learning techniques, particularly in classifying and identifying stars based on their features. This study focuses on the spectra-based classification of 11,408 B-type and 2422 hot subdwarf stars. The study employs baseline correction using Asymmetric Least Squares (ALS) to enhance classification accuracy. It applies the Pan-Core concept to identify 500 unique patterns or ranges for both types of stars. These patterns are the foundation for creating Support Vector Machine (SVM) models, including the linear (L-SVM), polynomial (P-SVM), and radial basis (R-SVM) kernels. Parameter tuning for the SVM models is achieved through cross-validation. Evaluation of the SVM models on test data reveals that the linear kernel SVM achieves the highest accuracy (87.0%), surpassing the polynomial kernel SVM (84.1%) and radial kernel SVM (80.1%). The average calibrated accuracy falls within the range of 90–95%. These results demonstrate the potential of using spectrum-based classification to aid astronomers in improving and expanding their understanding of stars, with a specific focus on the identification of hot subdwarf stars. This study presents a valuable investigation for astronomers, as it enables the classification of stars based on their spectra, leveraging machine learning techniques to enhance their knowledge and insights in astronomy.

Analysis of Factors Causing Urban Road Traffic Accidents Based on XGboost Algorithm

To further strengthen road traffic safety management and improve the accuracy of early warning systems for road traffic safety, a model for factors causing urban road traffic accidents based on the XGBoost algorithm is proposed. Firstly, SMOTE is used to process the unbalanced data, including supplementing the missing values, deleting the duplicate values, and visualizing the data. Then the prediction model is built by the XGBoost algorithm. By comparing and analyzing the results with the LR model, linear SVM model, DT model and Lightgbm model, the average accuracy rate of the XGBoost model reaches 0.979. Based on the XGBoost algorithm, the analysis of the factors causing urban road traffic accidents has better prediction performance, which can provide a reliable reference for preventing traffic accidents.

Prediction of breast cancer based on machine learning

Breast cancer is a frequent cancer that develops when normal cells in the breast transform into malignant cells. Breast cancer can arise from glandular tissue, muscular tissue, or fatty tissue in the breast. Many variables contribute to the risk of breast cancer, including genetics, environmental exposure, food, and lifestyle. Breast cancer should be detected early through breast self-examination, regular clinical evaluation, and mammography to identify any abnormal changes, In recent years, early detection of breast cancer in women has emerged as a beacon of hope and a pivotal point in the treatment of this dangerous disease, and its timely identification has become paramount. Modern advancements in technology, especially artificial intelligence algorithms, have played a vital role in developing systems that facilitate automated disease detection, diagnosis, rapid response, and a reduced risk of fatalities. This paper delves into a comparative study of various machine learning (ML) techniques, namely logistic regression (LR), support vector machines (SVM), linear SVM, Gaussian Naive Bayes (GNB), and artificial neural networks (ANNs). The evaluation metrics used in this study are accuracy and elapsed time. The results show that Gaussian Naive Bayes achieved the highest accuracy of 94.07% in just 0.005495 seconds, outperforming SVM (91.85%), linear SVM (90.19%), logistic regression (87.04%), and ANN (37.04%). These findings highlight the potential of Gaussian Naive Bayes in aiding the early detection of breast cancer, leading to more effective and timely interventions, ultimately improving patient outcomes. Keywords: Breast Cancer, Machine learning (ML), Logistic Regression (LR), Support Vector Machine (SVM), Linear SVM, Gaussian Naive Bayes (GNB) and Artificial Neural Networks (ANNs).

Prediction of the health state of lithium-ion power batteries using machine learning

Faced with the increasingly urgent issues of energy depletion and environmental pollution, various countries are actively supporting the new energy vehicle industry, making electric vehicle technology a focus of widespread attention among researchers. The Battery Management System (BMS) plays a crucial role in overseeing the power battery's operational parameters, with the estimation of the State of Health (SOH) being a pivotal function. Accurate estimation of SOH can improve the utilization efficiency and endurance of power batteries, extend their service life, and help users achieve the best balance between system safety and economic benefits. Machine learning methods provide a new solution to the SOH estimation problem. Without analyzing the complex aging mechanisms inside the battery, online SOH estimation can be completed by learning from historical aging data. In this study, the lithium battery dataset provided by NASA is used and trained through multiple linear regression models and support vector machine models to predict the SOH of lithium-ion batteries. The results demonstrate the accuracy and reliability of both methods in predicting SOH. Simultaneously, the prediction results of different feature values are compared to obtain the most accurate combination of feature values.

A Productivity Prediction Method of Fracture-Vuggy Reservoirs Based on the PSO-BP Neural Network

Reservoir productivity prediction is a key component of oil and gas field development, and the rapid and accurate evaluation of reservoir productivity plays an important role in evaluating oil field development potential and improving oil field development efficiency. Fracture-vuggy reservoirs are characterized by strong heterogeneity, complex distribution, and irregular development, causing great difficulties in the efficient prediction of fracture-vuggy reservoirs’ productivity. Therefore, a PSO-BP fracture-vuggy reservoir productivity prediction model optimized by feature optimization was proposed in this paper. The Chatterjee correlation coefficient was used to select the appropriate combination of seismic attributes as the input of the prediction model, and we applied the PSO-BP model to predict oil wells’ production in a typical fracture-vuggy reservoir area of Tahe Oilfield, China, with the selected seismic attributes and compared the accuracy with that provided by the BP neural network, linear support vector machine, and multiple linear regression. The prediction results using the four models based on the test set showed that compared with the other three models, the MSE of the PSO-BP model increased by 23% to 62%, the RMSE increased by 12 to 38 percent, the MAE increased by 18 to 44 percent, the SSE increased by 23 to 62 percent, and the R-square value increased by 2 to 13 percent. This comparison proves that the PSO-BP neural network model proposed in this paper is suitable for the productivity prediction of fracture-vuggy reservoirs and has better performance, which is of guiding significance for the development and production of fracture-vuggy reservoirs.

Unveiling intrusions: explainable SVM approaches for addressing encrypted Wi-Fi traffic in UAV networks

AbstractUnmanned aerial vehicles (UAVs), also known as drones, have become instrumental in various domains, including agriculture, geographic information systems, media, logistics, security, and defense. These UAVs often rely on wireless communication networks for data transmission, making them vulnerable to cyberattacks. To address these challenges, it is necessary to detect potential threats by analyzing the encrypted Wi-Fi traffic data generated by UAVs. This study aimed to develop a linear SVM model that is enhanced with explainable artificial intelligence (XAI) techniques and fine-tuned using Bayesian optimization for intrusion detection systems (IDSs); the model is specifically designed to identify malware threats targeting UAVs. This research utilized encrypted Wi-Fi traffic data derived from three different UAV networks, namely, Parrot Bebop 1, DBPower UDI, and DJI Spark, while considering unidirectional and bidirectional communication flow modes. SVM-based intrusion detection models have been modeled on these datasets, identified their key features using the local interpretable model-agnostic explanations (LIME) technique, and conducted a cost analysis of the proposed modeling approach. The incorporation of the LIME method enabled to highlight the features that are highly indicative of cyberattacks and provided valuable insights into the importance of each feature in the context of intrusion detection. In conclusion, this interpretable IDS model, fine-tuned with Bayesian optimization, demonstrated its superiority over the state-of-the-art methods, proving its efficacy in detecting and mitigating threats to UAVs while offering a cost-effective solution.