Grid Search Approach Research Articles

Mechanism of abrupt supershear rupture for the 2021 Maduo earthquake (Mw=7.4) in the northern Tibetan Plateau

The 2021 Maduo earthquake (Mw = 7.4) is the first sudden and abrupt supershear rupture ever documented in the history of seismology, in which subshear rupture changed to supershear within a short time and short distance. However, the mechanism for this special phenomenon remains unclear although previous workers have done much work on it. For this reason, we use the finite element method to study the effects of fault geometry near the epicenter of the earthquake on the generation of rupture scenarios and use the grid-search approach to find the optimal model. Our simulation results show that the special fault geometry with a curved bend near the epicenter in the eastern segment of the fault induced supershear rupture transition at ∼3–5 s after the rupture nucleation, leading to the abrupt and sudden occurrence of the supershear rupture in the earthquake, whereas the westward-going rupture behaved as subshear in general. Additionally, the modeling results suggest that the co-seismic slips along the fault are mainly controlled by unevenly distributed dynamic friction coefficients. The modeling results also suggest that the other complex geometry of the fault, such as stepovers, cannot encourage the abrupt supershear rupture in the Maduo earthquake. Therefore, in this work, we may provide a new perspective for the study of the dynamic mechanism of supershear rupture.

Analysis Grid Search Optimization of Machine Learning Models for Slope Stability Prediction Supports the Design Construction of Geotechnical Structures and Environmental Development

The accurate prediction of slope stability is crucial for the design and construction of geotechnical structures, as well as for environmental development and risk mitigation. This study explores the application of machine learning (ML) models optimized using the Grid Search method to enhance slope stability predictions. In this study, an in-depth analysis of seven prediction models Logistic Regression (LR), K-Nearest Neighbor (KNN), Support Vector Classifier (SVC), CatBoost, and RUSBoost is provided. These models were evaluated using a grid search approach to find the optimal hyperparameters. The model takes several input features, including pore water ratio (ru), height (H), unit weight (Ƴ), cohesion (c), slope angle (β), and angle of internal friction (ɸ). The output is the slope status, either stable (1) or unstable (0). The generalization ability of classification models is improved by using a 5-fold cross-validation (CV). Evaluation indicators such as AUC, accuracy, and kappa were analyzed, and CatBoost outperformed other machine learning models with the highest AUC of 0.823, accuracy of 0.874, and kappa of 0.642. The results indicate that CatBoost is a highly effective tool for predicting slope stability, surpassing other models in classification accuracy. The capacity and efficiency of RF in deformation prediction models make it the most accurate tool available for forecasting slope stability. Additionally, a comprehensive analysis of feature sensitivity was conducted to determine the most significant characteristics for predicting slope stability. These findings not only enhance geotechnical safety but also contribute to sustainable environmental development by preventing landslides, reducing soil erosion, and supporting responsible land-use planning. The integration of machine learning into slope stability analysis promotes ecological preservation and long-term resilience in infrastructure projects.

Small‐UAV GPS Spoofing Attack Detection and Multilabel Classification

ABSTRACTSmall unmanned aerial vehicles (UAVs) rely heavily on global positioning systems (GPS) for navigation. Nevertheless, it is possible to launch GPS spoofing attacks, which makes the UAVs' ability to track themselves inexplicable. To solve this issue, this study proposed an ensemble stacking method to identify GPS signal spoofing in small UAVs. It contributes to the dataset extraction that contains information from real GPS signals collected from various locations to simulate an automated and stationary autonomous vehicle using an accessible software radio peripheral unit established as a GPS receiver and 13 features extracted from eight parallel mediums. We use a grid search approach to identify and fine‐tune hyperparameters to improve the detection rate. The results show that the proposed approach performed well in traditional machine learning approaches and accurately predicts spoofing attacks with an accuracy of 98%. The results suggested that the proposed ensemble stacking performs better than alternative methods. The findings show that the proposed approach outperforms spoofing small‐UAV GPS signal diagnosis techniques.

Enhancing Chronic Pain Nursing Diagnosis Through Machine Learning: A Performance Evaluation.

This study proposes an evaluation of the efficacy of machine learning algorithms in classifying chronic pain based on Italian nursing notes, contributing to the integration of artificial intelligence tools in healthcare within an Italian linguistic context. The research aimed to validate the nursing diagnosis of chronic pain and explore the potential of artificial intelligence (AI) in enhancing clinical decision-making in Italian healthcare settings. Three machine learning algorithms-XGBoost, gradient boosting, and BERT-were optimized through a grid search approach to identify the most suitable hyperparameters for each model. Therefore, the performance of the algorithms was evaluated and compared using Cohen's κ coefficient. This statistical measure assesses the level of agreement between the predicted classifications and the actual data labels. Results demonstrated XGBoost's superior performance, whereas BERT showed potential in handling complex Italian language structures despite data volume and domain specificity limitations. The study highlights the importance of algorithm selection in clinical applications and the potential of machine learning in healthcare, specifically addressing the challenges of Italian medical language processing. This work contributes to the growing field of artificial intelligence in nursing, offering insights into the challenges and opportunities of implementing machine learning in Italian clinical practice. Future research could explore integrating multimodal data, combining text analysis with physiological signals and imaging data, to create more comprehensive and accurate chronic pain classification models tailored to the Italian healthcare system.

Global Model Selection via Solution Paths for Robust Support Vector Machine.

Robust support vector machine (RSVM) using ramp loss provides a better generalization performance than traditional support vector machine (SVM) using hinge loss. However, the good performance of RSVM heavily depends on the proper values of regularization parameter and ramp parameter. Traditional model selection technique with gird search has extremely high computational cost especially for fine-grained search. To address this challenging problem, in this paper, we first propose solution paths of RSVM (SPRSVM) based on the concave-convex procedure (CCCP) which can track the solutions of the non-convex RSVM with respect to regularization parameter and ramp parameter respectively. Specifically, we use incremental and decremental learning algorithms to deal with the Karush-Khun-Tucker violating samples in the process of tracking the solutions. Based on the solution paths of RSVM and the piecewise linearity of model function, we can compute the error paths of RSVM and find the values of regularization parameter and ramp parameter, respectively, which corresponds to the minimum cross validation error. We prove the finite convergence of SPRSVM and analyze the computational complexity of SPRSVM. Experimental results on a variety of benchmark datasets not only verify that our SPRSVM can globally search the regularization and ramp parameters respectively, but also show a huge reduction of computational time compared with the grid search approach.

Demand Management in Hybrid Locomotives Through Aggregated Models of Supercapacitors and Railway Units

Most European Union governments and numerous railway operators have announced plans to replace most of their diesel units by 2030–2040. However, a significant portion of the rail network remains non-electrified. In some cases, the proposed solution has been to close certain tracks, but this approach entails considerable societal costs for small cities and represents a loss of prior railway investments. Consequently, hybrid locomotives and multiple units (either new or refurbished) emerge as a viable solution during this transitional period to enhance energy efficiency and preserve services on these lines, particularly for freight operations. These hybrid units can operate on both electrified and non-electrified tracks and can also serve as “railway prosumers”, contributing to both storage and generation in fully or partially electrified areas. However, implementing these “prosumer tasks” faces challenges, such as the rapid power demand fluctuations during acceleration and the loss of energy recovery potential during braking in hybrid or fully electric units. These losses may also impact the overall power system. This paper presents an alternative approach to modeling double-layer capacitors (supercapacitors) combined with electrical equivalent models for lithium-ion batteries. The Differential Transformation Method (DTM) is used to solve the non-linear ordinary differential equations governing the supercapacitor model, while parameter optimization is achieved through a grid search approach, demonstrating high accuracy compared with laboratory trials. This framework highlights the potential of hybrid units, as illustrated through simulations that analyze storage sizing, energy management, increased energy recovery, and changes in unit performance. These models facilitate a pre-feasibility evaluation of energy storage systems for hybrid railway applications.

Coseismic slip distribution of the 2024 Noto Peninsula earthquake deduced from dense global navigation satellite system network and interferometric synthetic aperture radar data: effect of assumed dip angle

The Mw 7.5 Noto Peninsula earthquake, which occurred on January 1, 2024, was considerably hazardous to the peninsula and surrounding regions owing to a strong motion, large-scale crustal deformation, and subsequent tsunami. Significant surface displacement was observed by the dense global navigation satellite system (GNSS) stations, including universities and SoftBank corporation sites, and synthetic aperture radar (SAR). To estimate reliable coseismic slip distribution and its uncertainties for this event, we used the dense GNSS and the line-of-sight displacements from the SAR based on the Bayesian optimization framework. Considering the listric fault structure of this source fault, we validated the fault dip angles using the grid-search approach in the slip estimation. The acquired models indicated reverse fault motion, including a right-lateral slip component, and two slip peaks were estimated in the eastern and western regions of the fault in the central peninsula, independent of the assumed dip angles. These locations correspond to regions of significant uplift and westward displacement. The dip angle assumption affects the horizontal and vertical component of the calculated displacements: a higher dip angle model (≥ 45°) well reproduces vertical components of GNSS and synthetic aperture radar displacements, whereas a lower dip angle model (< 45°) well reproduces horizontal displacements. Overall, a fault dip of 45° is plausible, although it is not consistent with the listric structure suggested by the seismic reflection survey and the aftershock distribution below the central part of the peninsula. To test such a listric fault model, we conducted a coseismic slip estimation assuming a relatively high (60°) and low (25°) dip angles for the shallow and deep sections of the fault, respectively. Even in this case, we acquired a slip model similar to that of a plain fault, which reasonably reproduced each component of the surface displacements as well as the simple plane fault models. These results suggest that listric geometry is also acceptable for the source faults of this event, although the flat geometry similarly explains the observations.Graphical

Development of machine learning algorithms to predict viral load suppression among HIV patients in Conakry (Guinea).

Viral load (VL) suppression is key to ending the global HIV epidemic, and predicting it is critical for healthcare providers and people living with HIV (PLHIV). Traditional research has focused on statistical analysis, but machine learning (ML) is gradually influencing HIV clinical care. While ML has been used in various settings, there's a lack of research supporting antiretroviral therapy (ART) programs, especially in resource-limited settings like Guinea. This study aims to identify the most predictive variables of VL suppression and develop ML models for PLHIV in Conakry (Guinea). Anonymized data from HIV patients in eight Conakry health facilities were pre-processed, including variable recoding, record removal, missing value imputation, grouping small categories, creating dummy variables, and oversampling the smallest target class. Support vector machine (SVM), logistic regression (LR), naïve Bayes (NB), random forest (RF), and four stacked models were developed. Optimal parameters were determined through two cross-validation loops using a grid search approach. Sensitivity, specificity, predictive positive value (PPV), predictive negative value (PNV), F-score, and area under the curve (AUC) were computed on unseen data to assess model performance. RF was used to determine the most predictive variables. RF (94% F-score, 82% AUC) and NB (89% F-score, 82% AUC) were the most optimal models to detect VL suppression and non-suppression when applied to unseen data. The optimal parameters for RF were 1,000 estimators and no maximum depth (Random state = 40), and it identified Regimen schedule_6-Month, Duration on ART (months), Last ART CD4, Regimen schedule_Regular, and Last Pre-ART CD4 as top predictors for VL suppression. This study demonstrated the capability to predict VL suppression but has some limitations. The results are dependent on the quality of the data and are specific to the Guinea context and thus, there may be limitations with generalizability. Future studies may be to conduct a similar study in a different context and develop the most optimal model into an application that can be tested in a clinical context.

Establishing the pig as a translational animal model for neurodevelopment.

Within the last few decades, the domestic pig has emerged as an advantageous biomedical animal model due to a vast number of similarities in realms of development and neuroanatomical features. Even so, a major challenge remains in how to translate time between the pig and human. Previously, researchers have developed a Translating Mammalian Time model that estimates the timing of 95 neurodevelopmental events across 9 mammalian species. By identifying the timing of these various events, one can include an additional animal into the model and assign a unique species score to predict the post-conception day (PCD) that other events will occur. Our objective was to conduct a comprehensive literature review of pig neurodevelopmental events to enable chronological comparison to other mammalian species, including humans. A total of 30 neurodevelopmental events with corresponding PCDs were identified, that were then used to optimize the pig's species score using grid search and gradient descent approaches. Across both methods, the same species score of 2.157 was derived with a residual sum of squares of 4260.46. This species score places the domestic pig between the cat (1.808) and the macaque (2.255), thereby reinforcing the translational power of the pig comparable to non-human primates.

User Interface Bug Classification Model Using ML and NLP Techniques: A Comparative Performance Analysis of ML Models

Analyzing user interface (UI) bugs is an important step taken by testers and developers to assess the usability of the software product. UI bug classification helps in understanding the nature and cause of software failures. Manually classifying thousands of bugs is an inefficient and tedious job for both testers and developers. Objective of this research is to develop a classification model for the User Interface (UI) related bugs using supervised Machine Learning (ML) algorithms and Natural Language Processing (NLP) techniques. Also, to assess the effect of different sampling and feature vectorization techniques on the performance of ML algorithms. Classification is based upon ‘Summary’ feature of the bug report and utilizes six classifiers i.e., Gaussian Naïve Bayes (GNB), Multinomial Naïve Bayes (MNB), Logistic Regression (LR), Support Vector Machines (SVM), Random Forest (RF) and Gradient Boosting (GB). Dataset obtained is vectored using two vectorization techniques of NLP i.e., Bag of Words (BoW) and Term Frequency-Inverse Document Frequency (TF-IDF). ML models are trained after vectorization and data balancing. The models ' hyperparameter tuning (HT) has also been done using the grid search approach to improve their efficacy. This work provides a comparative performance analysis of ML techniques using Accuracy, Precision, Recall and F1 Score. Performance results showed that a UI bug classification model can be built by training a tuned SVM classifier using TF-IDF and SMOTE (Synthetic Minority Oversampling Techniques). SVM classifier provided the highest performance measure with Accuracy: 0.88, Precision: 0.86, Recall: 0.85 and F1: 0.85. Result also inferred that the performance of ML algorithms with TF-IDF is better than BoW in most cases. This work provides classification of bugs that are related to only the user interface. Also, the effect of two different feature extraction techniques and sampling techniques on algorithms were analyzed, adding novelty to the research work.

Detection of Hidden Low-Frequency Earthquakes in Southern Vancouver Island with Deep Learning

Low-frequency earthquakes (LFEs) are small-magnitude earthquakes that are depleted in high-frequency content relative to traditional earthquakes of the same magnitude. These events occur in conjunction with slow slip events (SSEs) and can be used to infer the space and time evolution of SSEs. However, because LFEs have weak signals, and the methods used to identify them are computationally expensive, LFEs are not routinely cataloged in most places. Here, we develop a deep-learning model that learns from an existing LFE catalog to detect LFEs in 14 years of continuous waveform data in southern Vancouver Island. The result shows significant increases in detection rates at individual stations. We associate the detections and locate them using a grid search approach in a 3D regional velocity model, resulting in over 1 million LFEs during the performing period. Our resulting catalog is consistent with a widely used tremor catalog during periods of large-magnitude SSEs. However, there are time periods where it registers far more LFEs than the tremor catalog. We highlight a 16-day period in May 2010, when our model detects nearly 3,000 LFEs, whereas the tremor catalog contains only one tremor detection in the same region. This suggests the possibility of hidden small-magnitude SSEs that are undetected by current approaches. Our approach improves the temporal and spatial resolution of the LFE activities and provides new opportunities to understand deep subduction zone processes in this region.

Ensemble learning using Gompertz function for leukemia classification

Ensemble learning using Gompertz function for leukemia classification

A method of developing ensemble classifiers for recognizing audio data of various nature

The work developed a method of building ensemble classifiers for recognizing audio data of various nature. The method is a tentative date and requires the following steps. In the first step, the input datasets are selected, which are transformed and divided into training and test samples, respectively. RAVDESS datasets are chosen for the task of audio emotion recognition, music genre recognition is performed on the GTZAN dataset. In the second step, the following seven classifiers were created and investigated as elements of ensemble classifiers: K Nearest Neighbors, Support Vector Machine, Random Forest, XGBoost, Multilayer Perceptron, Convolutional Neural Network and Long Short-term Memory. In the process of training elementary classifiers, the corresponding hyperparameters were adjusted using the Grid Search approach. In the third step, elementary classifiers were combined using the stacking ensemble method with such types of aggregation as soft voting, hard voting, soft voting using the GOMPERTZ function. All possible ensemble combinations starting with three elementary classifiers for recognizing audio emotions and music genres were tested. Therefore, the total number of ensembles studied in the work was 297. The research results for the problem of audio emotion classification showed that the accuracy of recognition according to the Accuracy metric of the best ensemble classifier is 8.1% higher than that of the best elementary classifier in its composition, which is based on on MLP, and according to the F1 metric, this indicator is 8% higher. For the task of recognizing music genres, the corresponding indicators are higher by 5.6 % and 5.2 %, respectively.

Efficient prediction for Blast Furnace Gas holder level using novel preprocessing techniques and weight correction strategy

Efficient prediction for Blast Furnace Gas holder level using novel preprocessing techniques and weight correction strategy

Feasibility of Identifying Shale Sweet Spots by Downhole Microseismic Imaging

Several studies suggest that shale sweet spots are likely associated with a low Poisson’s ratio in the shale layer. Compared with conventional geophysical techniques with active seismic data, it is straightforward and cost-effective to delineate the distribution of 3D Poisson’s ratios using microseismic data. In this study, an alternating method is proposed to determine microseismic event locations, 3D P-wave velocity, and Poisson’s ratio models with data recorded from downhole monitoring arrays. The method combines the improved 3D traveltime tomography, which inverts P and S arrivals for 3D P-wave velocity and Poisson’s ratio structures simultaneously, and a 3D grid search approach for event locations in an iterative fashion. The traveltime tomography directly inverts the Poisson’s ratio structure instead of calculating the Poisson’s ratios from P- and S-wave velocities (i.e., Vp and Vs) that are inverted by conventional traveltime tomography separately. The synthetic results and analysis suggest that the proposed method recovers the true Poisson’s ratio model reasonably. Additionally, we apply the method to a field dataset, which indicates that it may help delineate the reservoir structure and identify potential shale sweet spots.

Novel way to predict stock movements using multiple models and comprehensive analysis: leveraging voting meta-ensemble techniques

The research introduces a method for anticipating stock market patterns by combining machine learning techniques with analysis methods. Multiple machine learning algorithms were integrated to address the limitations of stock market forecasting models. Using web scraping techniques, data were gathered from the S&amp;P500 index over seven years, from September 5, 2016, to August 5, 2023. Companies like Microsoft Corporation (MSFT), Amazon.com Inc. (AMZN), JPMorgan Chase &amp; Co (JPM), and Tesla, Inc. (TSLA) were selected based on their inclusion in the S&amp;P 500 index. LR, RF, SVC, ADAB, and XGBC algorithms were applied as models by utilising optimisation using grid search and single algorithm approaches. Voting methods were employed to combine predictions from these models. The study employed rigorous statistical analyses, including the Kruskal-Wallis test to assess overall differences, followed by Pairwise Dunn’s Test with Bonferroni Correction for detailed algorithm comparisons. Additionally, Bootstrapping was utilised to calculate Confidence Intervals (CI) for robust estimation of algorithm performance. The methodology covered data collection, preprocessing, model training, and performance assessment. The outcomes indicate that the proposed approach accurately forecasts stock trends precisely and dependably. This study contributes to refining stock market prediction methodologies by introducing a strategy that enhances prediction accuracy while offering investors and financial professionals insights. Furthermore, assessing algorithm performance across metrics and companies highlights the versatility and effectiveness of machine-learning approaches in the fields.

Enhanced Binary Classification of Gait Disorders Using a Machine Learning Majority Voting Approach.

This study introduces a machine learning-based methodology for classifying healthy individuals and those with gait disorders, employing a merged data set from 'GaitRec' and 'Gutenberg.' Key gait features were extracted from the normalized ground reaction force data that had 2435 subjects, including maximum peaks, load rate, lowest troughs, and area under the curve. Numerous machine learning models, such as Support Vector Machine, Logistic Regression, Random Forest, Gradient Boosting, K-Nearest Neighbor, Bagging, Adaboost, and Neural Network, were developed, and their hyperparameters were optimized using a grid search approach. The model's performance was evaluated using metrics such as accuracy, sensitivity, specificity, and F1 score. These assessment metrics are calculated through a repeated hold-out strategy, where 80% of the data is utilized for training the model, and the remaining 20% is reserved for testing purposes. Notably, the study highlights the superiority of an ensemble model that combines these algorithms through majority voting. This majority voting model, which was compared against individual models, achieved an accuracy of 96.63%, surpassing existing benchmarks in the field. This approach underscores the efficacy of ensemble techniques, particularly majority voting, in enhancing classification accuracy, thus contributing significantly to the field of biomedical signal processing by offering a scalable and cost-effective solution for accurate gait disorder identification.

A data-driven analysis to determine the optimal number of topics 'K' for latent Dirichlet allocation model

Topic modeling is an unsupervised machine learning technique successfully used to classify and retrieve textual data. However, the performance of topic models is sensitive to selecting optimal hyperparameters, the number of topics 'K' and Dirichlet priors 'α' and 'β.' This data-driven analysis aims to determine the optimum number of topics, 'K,' within the latent Dirichlet allocation (LDA) model. This work utilizes three datasets, namely 20-Newsgroups news articles, Wikipedia articles, and Web of Science containing science articles, to assess and compare various 'K' values through the grid search approach. The grid search approach finds the best combination of hyperparameter values by trying all possible combinations to see which performs best. This research seeks to identify the 'K' that optimizes topic relevance, coherence, and model performance by leveraging statistical metrics, such as coherence scores, perplexity, and topic distribution quality. Through empirical analysis and rigorous evaluation, this work provides valuable insights for determining the ideal 'K' for LDA models.

Cloud-based machine learning algorithms for anomalies detection

Gradient boosting machines harnesses the inherent capabilities of decision trees and meticulously corrects their errors in a sequential fashion, culminating in remarkably precise predictions. Word2Vec, a prominent word embedding technique, occupies a pivotal role in natural language processing (NLP) tasks. Its proficiency lies in capturing intricate semantic relationships among words, thereby facilitating applications such as sentiment analysis, document classification, and machine translation to discern subtle nuances present in textual data. Bayesian networks introduce probabilistic modeling capabilities, predominantly in contexts marked by uncertainty. Their versatile applications encompass risk assessment, fault diagnosis, and recommendation systems. Gated recurrent units (GRU), a variant of recurrent neural networks, emerges as a formidable asset in modeling sequential data. Both training and testing are crucial to the success of an intrusion detection system (IDS). During the training phase, several models are created, each of which can recognize typical from anomalous patterns within a given dataset. To acquire passwords and credit card details, "phishing" usually entails impersonating a trusted company. Predictions of student performance on academic tasks are improved by hyper parameter optimization of the gradient boosting regression tree using the grid search approach.

Enhancing tribological performance of hybrid fiber-reinforced composites through machine learning and response surface methodology

This study delves into the significant effects of sodium hydroxide (NaOH) treatment on the tribological properties of hybrid fiber-reinforced composites, specifically focusing on the combination of paddy straw (PS) and pineapple leaf (PALF) in a polyester matrix. By leveraging Artificial Neural Networks (ANNs) to predict the Specific Wear Rate (SWR) and Coefficient of Friction (COF), the research employs a grid search approach for hyperparameter optimization. This optimization process results in an optimal ANN architecture with impressive accuracy, showcasing low mean absolute error and high R-squared values of 0.991 and 0.986 for SWR and COF predictions, respectively. Utilizing the Design of Experiments (DOE), the study systematically analyzes the intricate interplay of disc speed, wear duration, and NaOH treatment percentage, with a specific focus on SWR and COF as pivotal tribological metrics. The Analysis of Variance (ANOVA) results underscore the substantial impact of duration and treatment percentage on wear characteristics. Additionally, quadratic regression models reveal nuanced correlations, highlighting the sensitivity of SWR to NaOH percentage and the influence of disc speed, duration, and treatment percentage on COF. This outcome emphasizes the efficacy of these parameters in achieving superior tribological performance in hybrid composites. Beyond contributing to a profound understanding of wear characteristics, this work introduces an innovative dimension through optimized ANN modeling, ensuring a more accurate and fine-tuned predictive model.