Performance Of Model Research Articles

Analysis of language identification algorithms for regional Indonesian languages

Detecting local languages in Indonesia is essential for recognizing linguistic diversity, promoting intercultural understanding, preserving endangered languages, and improving access to education and services. By identifying and documenting these languages, we can support language preservation efforts, provide tailored resources for communities, and celebrate the unique cultural heritage of different ethnic groups. Ultimately, this encourages a more accepting and open-minded society, prioritizing various languages and cultural customs. This research aims to identify the most suitable algorithm for language detection in Indonesian regional languages and gain insights into their unique characteristics through n-gram analysis. By understanding language diversity, the study contributes to preserving Indonesia's cultural and linguistic heritage and improving language detection techniques. This study compares the performance of five algorithms (Naïve Bayes, K-nearest neighbors (KNN), least-squares, Kullback Leibler divergence, and Kolmogorov Smirnov test) to determine the most accurate and efficient method for language identification. Incorporating trigram features alongside unigrams and bigrams significantly improved the model's performance, with F1 scores increasing from 0.923 to 0.959. The study found that using more features leads to better accuracy, with Naïve Bayes and KNN emerging as the top-performing algorithms for language identification.

Steady state stability assessment of voltage source converter in a DC microgrid under weak grid conditions

Traditionally, steady-state assessment involves analyzing numerous variables using Eigen analysis. This paper presents a decision support application for diagnosing the steady-state assessment of droop-controlled voltage source inverters in islanded microgrid operations or weak grid operations with reduced input attributes. This paper proposes an approach using feature extraction from the state space variables of the droop-controlled voltage source inverter (VSI). Photovoltaic (PV) and wind energy sources are considered with their stipulated power-delivering capability considered. To improve the generalization of the predictive model, preprocessing techniques are employed to eliminate data distortions. Dimensionality reduction is achieved through principal component analysis (PCA) applied to the steady-state variables. The evaluation of the VSI's steady-state stability is conducted utilizing support vector classification algorithm. To ascertain the reliability of the steady-state stability classification, an assessment of the support vector machine (SVM) model's performance is carried out, which includes the examination of metrics like the area under the curve (AUC) and the receiver operating characteristics (ROC) curve. The findings from the assessment of VSI's steady-state stability indicate a commendable level of performance, achieving an accuracy rate of 93.5%.

Detection and Quantification of 5moU RNA Modification from Direct RNA Sequencing Data

Background: Chemically modified therapeutic mRNAs have gained momentum recently. In addition to commonly used modifications (e.g., pseudouridine), 5moU is considered a promising substitution for uridine in therapeutic mRNAs. Accurate identification of 5-methoxyuridine (5moU) would be crucial for the study and quality control of relevant in vitro-transcribed (IVT) mRNAs. However, current methods exhibit deficiencies in providing quantitative methodologies for detecting such modification. Utilizing the capabilities of Oxford nanopore direct RNA sequencing, in this study, we present NanoML-5moU, a machine-learning framework designed specifically for the read-level detection and quantification of 5moU modification for IVT data. Materials and Methods: Nanopore direct RNA sequencing data from both 5moU-modified and unmodified control samples were collected. Subsequently, a comprehensive analysis and modeling of signal event characteristics (mean, median current intensities, standard deviations, and dwell times) were performed. Furthermore, classical machine learning algorithms, notably the Support Vector Machine (SVM), Random Forest (RF), and XGBoost were employed to discern 5moU modifications within NNUNN (where N represents A, C, U, or G) 5-mers. Result: Notably, the signal event attributes pertaining to each constituent base of the NNUNN 5- mers, in conjunction with the utilization of the XGBoost algorithm, exhibited remarkable performance levels (with a maximum AUROC of 0.9567 in the "AGTTC" reference 5-mer dataset and a minimum AUROC of 0.8113 in the "TGTGC" reference 5-mer dataset). This accomplishment markedly exceeded the efficacy of the prevailing background error comparison model (ELIGOs AUC 0.751 for sitelevel prediction). The model's performance was further validated through a series of curated datasets, which featured customized modification ratios designed to emulate broader data patterns, demonstrating its general applicability in quality control of IVT mRNA vaccines. The NanoML-5moU framework is publicly available on GitHub (https://github.com/JiayiLi21/NanoML-5moU). Conclusion: NanoML-5moU enables accurate read-level profiling of 5moU modification with nanopore direct RNA-sequencing, which is a powerful tool specialized in unveiling signal patterns in in vitro-transcribed (IVT) mRNAs.

Transfer learning for anatomical structure segmentation in otorhinolaryngology microsurgery.

Reducing the annotation burden is an active and meaningful area of artificial intelligence (AI) research. Multiple datasets for the segmentation of two landmarks were constructed based on 41257 labelled images and 6 different microsurgical scenarios. These datasets were trained using the multi-stage transfer learning (TL) methodology. The multi-stage TL enhanced segmentation performance over baseline (mIOU 0.6892 vs. 0.8869). Besides, Convolutional Neural Networks (CNNs) achieved a robust performance (mIOU 0.8917 vs. 0.8603) even when the training dataset size was reduced from 90% (30078 images) to 10% (3342 images). When directly applying the weight from one certain surgical scenario to recognise the same target in images of other scenarios without training, CNNs still obtained an optimal mIOU of 0.6190±0.0789. Model performance can be improved with TL in datasets with reduced size and increased complexity. It is feasible for data-based domain adaptation among different microsurgical fields.

Low-resolution facial emotion recognition on low-cost devices

The low-resolution input image is a crucial challenge for applying facial emotion recognition in real-world scenarios. The critical problem is that valuable object features are relatively lost in the extraction process due to their small size. On the other hand, this vision system is required by a machine to run smoothly on low-cost devices. Facial emotion recognition using a lightweight feature extractor is proposed in this study to effectively capture crucial facial components in a low-resolution image. To compromise the running speed, this work offers an efficient feature convolution to discriminate specific facial features. In addition, the system is embedded with an attentive module to capture important features and correlate them. Our model performance is evaluated on low-resolution public datasets achieving the accuracy of 97.34\%, 81.10\%, and 80.12\% on KDEF, RFDB, and FER-plus, respectively. The practical application demands that the deep learning model can operate fast on inexpensive devices. Consequently, the model achieved a speed of 290 FPS on a CPU device.

Ensemble methods for quantification of potassium oxide in ChemCam Mars and laboratory spectra

In this paper we test new approaches for predicting the amount of element oxides in rock samples from the ChemCam instrument suite onboard the NASA Curiosity rover by focusing on K2O. Using the expanded dataset compiled by Gasda et al. (2021) with and without the Earth to Mars (E2M and NoE2M) transformation discussed in Clegg et al. (2017) we trained blended submodels using the “double blending” technique and compared these to ensemble methods (Random Forest, ExtraTrees, and Gradient Boosting Regression). We found that ensemble methods performed similar to blended submodels when looking at RMSE-P on the laboratory spectra and provided significant advantages when looking at spectra coming from Mars. For the full model, blended submodels achieved an RMSE-P of 0.62 and 0.60 (E2M and NoE2M respectively) while Gradient Boosting Regression resulted in a slightly improved RMSE-P of 0.59 and 0.60. More importantly, by employing a local RMSE-P estimation technique where model performance is evaluated based on nearby test samples we found that using ensemble methods can lower the quantification limit for K2O from the current value of ≈0.6 wt% to ≈0.08 wt% using Extra Trees and Random Forest. This would allow for a much larger range of K2O values to be quantified on Mars with greater certainty given that most targets seen on Mars tend to have <1 wt% K2O. Finally, we used both Mean Decrease in Impurity (MDI) and permutation importance techniques to investigate the wavelengths used by the ensemble methods and found that they correspond to known potassium emission lines. This suggests that ensemble methods can provide an easier to train and improved alternative to blended submodels for predicting potassium compositions from Laser Induced Breakdown Spectroscopy (LIBS) data.

From drought hazard to risk: A spring wheat vulnerability assessment in the Canadian Prairies

The Canadian drought monitor (CDM) provides monthly assessments of drought severity but does not provide guidance on the potential agricultural impacts. This study extends the CDM to a comprehensive risk assessment framework by assessing the vulnerability of spring wheat to drought timing and severity in the Canadian Prairies. To achieve this, a bayesian neural network (BNN) model was developed that predicts high-resolution wheat yields and uncertainties using the CDM and normalized difference vegetation index (NDVI) data from 2010 to 2021. Model performance is comparable to previous studies, with errors ranging from 13% to 18% of mean yields and R2 values from 0.43 to 0.70 that peak in early August. The model was then applied under various drought conditions to evaluate how different drought timings and severities affect yield vulnerability. The results demonstrate that drought exposure from June to August reduces wheat yields in proportion to drought severity while abnormally dry conditions may have a positive impact of yield, particularly in August. Yield uncertainty decreases substantially from June to July while yield prediction accuracy increases, suggesting that mitigation efforts are most effective prior to July. The final component of the research involved creating operational risk maps that integrate the yield vulnerability assessment with the CDM. These maps were applied in a case study to demonstrate their utility in providing actionable, real-time information for agricultural stakeholders. Overall, this study contributes to the body of knowledge regarding crop sensitivities to varying drought conditions and represents a methodological advancement by enabling high-resolution, operational agricultural risk assessments at scale. By integrating detailed vulnerability data with drought hazard monitoring from the CDM, the information provided by these risk maps can be leveraged by agricultural stakeholders in near real-time to make well-informed decisions that mitigate the impacts of drought on wheat production, enhancing the resilience of agricultural systems to climatic variability.

Fine tuning attribute weighted naïve Bayes model for detecting anxiety disorder levels of online gamers

This research applies the fine tuning attribute weighted naïve Bayes (FTAWNB) model using ordinal data. It is known that in previous research, the FTAWNB model outperformed its competitors on the dataset used. However, the FTAWNB model has not been applied in the mental health domain that uses ordinal data. Therefore, this research used the anxiety gamers dataset to test the fine-tuning attribute weighted Naïve Bayes (FTAWNB) model. Anxiety disorders are mental health disorders that can indicate the emergence of a gaming disorder. Gamers can experience anxiety disorders classified into four classes, namely minimal, mild, moderate, and severe anxiety. Then compare the results by FTAWNB obtained with three other naïve Bayes algorithms, namely Gaussian naïve Bayes, multinomial naïve Bayes, and categorical naïve Bayes, using the same dataset. Model performance is measured based on accuracy, precision, recall, and processing time. The test results show that the FTAWNB outperforms the other three models' accuracy, precision, and recall, with an accuracy value of 99.22%. While the accuracy of Gaussian NB is 91.132%, Categorical is 91.592%, and multinomial naïve Bayes is 61.104%. However, the FTAWNB takes slightly longer than the other three models' processing time. The FTAWNB takes 0.07 seconds to build the model and 0.05 seconds to test the model on training data.

Clinical Variables and Radiomics Features for Predicting Pneumothorax in Patients Undergoing CT-guided Transthoracic Core Needle Biopsy.

Purpose To develop a prediction model combining both clinical and CT texture analysis radiomics features for predicting pneumothorax complications in patients undergoing CT-guided core needle biopsy. Materials and Methods A total of 424 patients (mean age, 65.6 years ± 12.7 [SD]; 232 male, 192 female) who underwent CT-guided core needle biopsy between January 2021 and October 2022 were retrospectively included as the training data set. Clinical and procedure-related characteristics were documented. Texture analysis radiomics features were extracted from the subpleural lung parenchyma traversed by needle. Moderate pneumothorax was defined as a postprocedure air rim of 2 cm or greater. The prediction model was developed using logistic regression with backward elimination, presented by linear fusion of the selected features weighted by their coefficients. Model performance was assessed using the area under the receiver operating characteristic curve (AUC). Validation was conducted in an external cohort (n = 45; mean age, 58.2 years ± 12.7; 19 male, 26 female) from a different hospital. Results Moderate pneumothorax occurred in 12.0% (51 of 424) of the training cohort and 8.9% (four of 45) of the external test cohort. Patients with emphysema (P < .001) or a longer needle path length (P = .01) exhibited a higher incidence of moderate pneumothorax in the training cohort. Texture analysis features, including gray-level co-occurrence matrix cluster shade (P < .001), gray-level run-length matrix low gray-level run emphasis (P = .049), gray-level run-length matrix run entropy (P = .003), gray-level size-zone matrix gray-level variance (P < .001), and neighboring gray-tone difference matrix complexity (P < .001), showed higher values in patients with moderate pneumothorax. The combined clinical-radiomics model demonstrated satisfactory performance in both the training (AUC 0.78, accuracy = 71.9%) and external test cohorts (AUC 0.86, accuracy 73.3%). Conclusion The model integrating both clinical and radiomics features offered practical diagnostic performance and accuracy for predicting moderate pneumothorax in patients undergoing CT-guided core needle biopsy. Keywords: Biopsy/Needle Aspiration, Thorax, CT, Pneumothorax, Core Needle Biopsy, Texture Analysis, Radiomics, CT Supplemental material is available for this article. © RSNA, 2024.

Machine learning prediction of refractory ventricular fibrillation in out-of-hospital cardiac arrest using features available to EMS

BackgroundShock-refractory ventricular fibrillation (VF) or ventricular tachycardia (VT) is a treatment challenge in out-of-hospital cardiac arrest (OHCA). This study aimed to develop and validate machine learning models that could be implemented by emergency medical services (EMS) to predict refractory VF/VT in OHCA patients. MethodsThis was a retrospective study examining adult non-traumatic OHCA patients brought into the emergency department by Singapore EMS from the Pan-Asian Resuscitation Outcomes Study (PAROS) registry. Data from April 2010 to March 2020 were extracted for this study. Refractory VF/VT was defined as VF/VT persisting or recurring after at least one shock. Features were selected based on expert clinical opinion and availability to dispatch prior to arrival at scene. Multivariable logistic regression (MVR), LASSO and random forest (RF) models were investigated. Model performance was evaluated using receiver operator characteristic (ROC) area under curve (AUC) analysis and calibration plots. Results20,713 patients were included in this study, of which 860 (4.1%) fulfilled the criteria for refractory VF/VT. All models performed comparably and were moderately well-calibrated. ROC-AUC were 0.732 (95% CI, 0.695 – 0.769) for MVR, 0.738 (95% CI, 0.701 – 0.774) for LASSO, and 0.731 (95% CI, 0.690 – 0.773) for RF. The shared important predictors across all models included male gender and public location. ConclusionThe machine learning models developed have potential clinical utility to improve outcomes in cases of refractory VF/VT OHCA. Prediction of refractory VF/VT prior to arrival at patient’s side may allow for increased options for intervention both by EMS and tertiary care centres.

Enhancing currency prediction in international e-commerce: Bayesian-optimized random forest approach using the Klarna dataset

In the ever-evolving landscape of global commerce, marked by the convergence of digital transformation and borderless markets, this research addresses the intricate challenges of currency exchange and risk management. Leveraging Bayesian optimization, the study fine-tunes the random forest algorithm using the extensive Klarna E-commerce dataset. Through systematic analysis, the research uncovers insights into managing currency prediction amid dynamic global markets. Emphasizing the role of Bayesian optimization parameters, the study reveals nuanced trade-offs in model performance. Notably, the optimal simulation, conducted with 14 iterations, 1 job, and a random state set to 684, exhibits a standout performance, showcasing a negative mean squared error (MSE) of approximately -0.9891 and an accuracy rate of 74.63%. The primary objective is to assess the impact of Bayesian optimization in enhancing the random forest algorithm's predictive capabilities, particularly in currency prediction within international e-commerce. These findings offer refined strategies for businesses navigating the intricate landscape of global finance, empowering decision-making through a comprehensive understanding of data, algorithms, and challenges in international commerce.

Using natural language processing to evaluate the impact of specialized transformers models on medical domain tasks

We are presently living in the age of intelligent machines, machines are rapidly imitating humans as a result of technological breakthroughs and advances in machine learning, deep learning, and artificial intelligence. In our work, we based our approach on the idea of utilizing a specialized corpus to enhance the performance of a pre-trained language model. We utilized the following approach: (V = vocabulary domain, C1 = initial corpus, C2 = specialization corpus). We applied this approach with different combinations such as (V = general, C1 = general, C2 = ∅), (V = general, C1 = general, C2 = medical), (V = medical, C1 = medical, C2 = ∅), and (V = medical, C1 = medical, C2 = medical) to compare the performance of a general bidirectional encoder representations from transformers model and specialized BERT models for the medical domain. In addition, we evaluated the model’s using informatics for integrating biology and the bedside, and drug-drug interaction datasets to measure their effectiveness in medical tasks.

Hybrid model for extractive single document summarization: utilizing BERTopic and BERT model

Extractive text summarization has been a popular research area for many years. The goal of this task is to generate a compact and coherent summary of a given document, preserving the most important information. However, current extractive summarization methods still face several challenges such as semantic drift, repetition, redundancy, and lack of coherence. A novel approach is presented in this paper to improve the performance of an extractive summarization model based on bidirectional encoder representations from transformers (BERT) by incorporating topic modeling using the BERTopic model. Our method first utilizes BERTopic to identify the dominant topics in a document and then employs a BERT-based deep neural network to extract the most salient sentences related to those topics. Our experiments on the cable news network (CNN)/daily mail dataset demonstrate that our proposed method outperforms state-of-the-art BERT-based extractive summarization models in terms of recall-oriented understudy for gisting evaluation (ROUGE) scores, which resulted in an increase of 32.53% of ROUGE-1, 47.55% of ROUGE-2, and 16.63% of ROUGE-L when compared to baseline BERT-based extractive summarization models. This paper contributes to the field of extractive text summarization, highlights the potential of topic modeling in improving summarization results, and provides a new direction for future research.

Hybrid machine learning for stock price prediction in the Moroccan banking sector

Analyzing historical stock market data using machine-learning techniques is crucial for data scientists and researchers to optimize stock price prediction models. This study uses machine learning regression algorithms and feature selection methods to optimize a simulated stock price prediction model using real historical data from Bank of Africa, a Moroccan bank. The approach compares multiple supervised regression algorithms, such as linear regression, extreme gradient boosting, ordinary least squared, random forest regressor, a linear least-squares L2-regularized, epsilon-support vector regression, and linear support vector regression. Each of these algorithms is associated with different feature selection algorithms to improve the performance of the prediction model. The analysis results revealed that hybridizing algorithms between the highest score percentiles, univariate linear regression, and linear support vector regression perform better according to the root mean squared error and R2-Score measures. This approach overcomes the problems associated with high-dimensional data by reducing the number of features and improving prediction accuracy.

Performance of machine learning models to forecast PM10 levels

Machine learning techniques have garnered considerable attention in modern technologies due to their promising outcomes across various domains. This paper presents the comprehensive methodology of an optimized and efficient forecasting approach for Particulate Matter 10, specifically tailored to predefined locations. The execution of a comparative analysis involving eight models enables the identification of the most suitable model that aligns with the primary research objective. Notably, the test results underscore the superior performance of an ensemble model, which integrates state-of-the-art methodologies, surpassing the performance of the other seven state-of-the-art models. Adopting a case-specific methodology with machine learning techniques contributes to achieving a notably high regression coefficient (R²≈1) across all models. Furthermore, the study underscores the potential for future endeavors in predicting location-specific environmental factors.•This study focused on forecasting PM10 with machine learning models with the consideration of air quality factors and meteorological factors•Ensemble model was developed for the forecasting purposes with higher performance.

Sensitivity of global hydrological models to potential evapotranspiration estimation methods in the Senegal River Basin (West Africa)

Study regionSenegal River Basin in West Africa Study focusThis paper aims to evaluate the sensitivity of global hydrological models to potential evapotranspiration (PET) methods in the Senegal River Basin. Potential evapotranspiration is estimated using 21 methods and its influence on the performance of three GR models (GR4J, GR5J and GR6J) is investigated in five catchments. The data used are mean rainfall, discharge and observed and calculated PET over the period 1984-1995. PET is calculated based on observed climate data and those from NASA POWER reanalysis data. The methodology consists in: (i) comparing the consistency of reanalysis data with respect to the observed PET, (ii) assessing the robustness of GR models and their sensitivity to different PET estimation methods. The evaluation criteria used to assess the performance of the hydrological model are KGE and PBIAS. New hydrological insights for the regionGood consistency is obtained between PET calculated with observed and reanalysis data. The GR4J and GR5J are more efficient to simulate flows in the Senegal River sub-basins. The aerodynamic PET methods perform well with the three hydrological models. However, in this context where data are scarce, temperature methods such as Droogers and Allen are a good choice for hydrological modeling. The results also show that the GR models have the ability to adapt to poorly estimated PET.

Real-time phishing detection using deep learning methods by extensions

Phishing is an attack method that relies on a user’s insufficient vigilance and understanding of the internet. For example, an attacker creates an online transaction website and tricks users into logging into the fake website to steal their personal information, such as credit card numbers, email addresses, phone numbers, and physical addresses. This paper proposes implementing an extension to prevent phishing for internet users. In particular, this study develops a smart warning feature for the proposed extension using deep learning models. The proposed extension installed in the web browser protects users by checking for, warning about, and preventing untrusted connections. This study evaluated and compared the performance of machine learning models using a malicious uniform resource locator (URL) dataset containing 651,191 data samples. The results of the investigation confirm that the proposed extension using a convolutional neural network (CNN) achieved a high accuracy of 98.4%.

An intelligent deep residual learning framework for tomato plant leaf disease classification

Modern agriculture has been fascinated by various advancements in agriculture and the processing of foods and supply management that provision farmers to improve production. The health of plants is essential to improve production and economic growth. Diseases in plants can affect production and create a rigorous impact on the quality and create a hazard to food safety. Hence, detecting and classifying plant leaf diseases is essential to prevent the disease spread across the plants in the agriculture field and to improve productivity. The researchers in existing frameworks utilized artificial intelligence and machine learning techniques to demonstrate noteworthy solutions. However, a few issues exist related to noises in the images, hyperparameter selection problems, and over-fitting problems that influence prediction accuracy. The proposed model jellyfish ResNet(JF-ResNet) works well to achieve a better accuracy level by incorporating jellyfish optimized ResNET for tomato plant leaf disease identification and classification. The performance metrics such as Accuracy, specificity, sensitivity, and F1-Score is used to evaluate the performance of the JF-ResNet model. The proposed model achieves 97.3% accuracy, 95.3% sensitivity, 96.1% specificity, 96.9% recall, 96.4% precision and 97.1% F1-Score.

Multi-granularity tooth analysis via YOLO-based object detection models for effective tooth detection and classification

Accurate detection and classification of teeth is the first step in dental disease diagnosis. However, the same class of tooth exhibits significant variations in surface appearance. Moreover, the complex geometrical structure poses challenges in learning discriminative features among the different tooth classes. Due to these complex features, tooth classification is one of the challenging research domains in deep learning. To address the aforementioned issues, the presented study proposes discriminative local feature extraction at different granular levels using YOLO models. However, this necessitates a granular intra-oral image dataset. To facilitate this requirement, a dataset at three granular levels (two, four, and seven teeth classes) is developed. YOLOv5, YOLOv6, and YOLOv7 models were trained using 2,790 images. The results indicate superior performance of YOLOv6 for two-class classification problems. The model generated a mean average precision (mAP) value of 94%. However, as the granularity level is increased, the performance of YOLO models decreases. For, four and seven-class classification problems, the highest mAP value of 87% and 79% was achieved by YOLOv5 respectively. The results indicate that different levels of granularity play an important role in tooth detection and classification. The YOLO’s performance gradually decreased as the granularity decreased especially at the finest granular level.

River Stream Flow Prediction through Advanced Machine Learning Models for Enhanced Accuracy

The Narmada River basin, a significant water resource in central India, plays a crucial role in supporting agricultural, industrial, and domestic water supply. Effective management of this basin requires accurate streamflow forecasting, which has become increasingly important. This study delves into streamflow forecasting using historical data from five major river stations, covering the upper reaches in the East and middle sections. The dataset spans from 1978 to 2020 and undergoes rigorous screening and preparation, including normalization using StandardScaler. The research adopts a comprehensive approach, developing models for training on 70% of historical data, validation on the most current 15%, and testing against future targets with another 15% of the data. To achieve precise predictions, a suite of machine learning models is employed, including CatBoost, LGBM (Light Gradient Boosting Machine), Random Forest, and XGBoost. Performance evaluation of these models relies on key indices such as mean squared error (MSE), mean absolute error (MAE), root mean square error (RMSE), root mean square percent error (RMSPE), normalized root mean squared error (NRMSE), and R-squared (R2). Notably, among these models, Random Forest emerges as the most robust for streamflow prediction, showcasing its effectiveness in handling the complexities of hydrological forecasting. This research contributes significantly to the field of streamflow forecasting in the Narmada River basin by providing insights into the performance of various machine learning models. The findings not only enhance our understanding of watershed dynamics but also highlight the pivotal role that machine learning can play in improving hydrological forecasting for sustainable watershed management.