Multiple Evaluation Metrics Research Articles

What Is Your Favorite Gender, MLM? Gender Bias Evaluation in Multilingual Masked Language Models

Bias is a disproportionate prejudice in favor of one side against another. Due to the success of transformer-based masked language models (MLMs) and their impact on many NLP tasks, a systematic evaluation of bias in these models is now needed more than ever. While many studies have evaluated gender bias in English MLMs, only a few have explored gender bias in other languages. This paper proposes a multilingual approach to estimating gender bias in MLMs from five languages: Chinese, English, German, Portuguese, and Spanish. Unlike previous work, our approach does not depend on parallel corpora coupled with English to detect gender bias in other languages using multilingual lexicons. Moreover, a novel model-based method is presented to generate sentence pairs for a more robust analysis of gender bias. For each language, lexicon-based and model-based methods are applied to create two datasets, which are used to evaluate gender bias in an MLM specifically trained for that language using one existing and three new scoring metrics. Our results show that the previous approach is data-sensitive and unstable, suggesting that gender bias should be assessed on a large dataset using multiple evaluation metrics for best practice.

A large-scale riverbank erosion risk assessment model integrating multi-source data and explainable artificial intelligence (XAI)

The impact of riverbank erosion poses serious threat to the environment, socio-economics and human safety. Due to the extremely complex mechanisms of erosion, assessing the risk of riverbank erosion is challenging. To address this, we propose an interpretable intelligent model framework to accurately assess large-scale riverbank erosion risk. Firstly, we constructed a multi-source dataset that encompasses 29 riverbank erosion influencing factors. Subsequently, by employing an adaptive feature weighting method, a comprehensive water level factor was synthesized, unifying data dimensions. The Relief algorithm was used to identify influential features for riverbank erosion, and an adaptive feature weighting SMOTE (AFW-SMOTE) algorithm was developed to balance the riverbank erosion dataset. Additionally, an ELM and BiGRU autoencoder was designed to effectively capture and learn key information from static and dynamic features. Finally, the outputs of the two autoencoders were integrated using the XGBoost algorithm to produce riverbank erosion risk assessment results, and the risks were visualized. This model not only performs excellently across multiple evaluation metrics but also significantly surpasses 22 other machine learning models. By integrating the Shapley value method, it enhances the model’s interpretability. This provides policymakers and relevant environmental management agencies with a powerful tool to scientifically assess and manage the risk of riverbank erosion.

Comprehensive evaluation of the distribution, transport and ecological risk of heavy metals in intra-urban river sediments using high-resolution techniques

Determining the distribution trends, transport mechanisms, and ecological risks of heavy metals (HMs) in urban river sediments is essential for the government to conduct appropriate remediation work. In this study, we collected sediment cores from the Yayao Waterway in Foshan City, China. The vertical distribution profiles of dissolved and labile Fe, Mn, Cd, Zn, Cu, Cr, Ni, Pb, As, and Co in the sediments were obtained using the thin-film diffusive gradient (DGT) and high-resolution peeper (HR-Peeper) techniques. In addition, the transport rates, contamination levels, and ecological concerns of the HMs were evaluated using the European Community Bureau of Reference (BCR) sequential extraction technique, the DGT-induced sediment fluxes (DIFS) model, and multiple contamination evaluation metrics. The results showed that most of the DGT-labile HMs were associated with Fe/Mn (hydrogen) oxides, and in particular, Zn, Ni, and Cr showed a significant negative correlation with Fe/Mn (p < 0.001). Additionally, Cd had the highest bioavailability (89.17%), and its net diffusive flux at the sediment–water interface (SWI) was positive, which indicated a high release risk from the sediment. However, the R-value of Cd based on the DGT-induced sediment fluxes (DIFS) operation was extremely low, suggesting that although Cd had the biggest supply pool of releases, its release rate was slow. The majority of sampling sites had significantly higher total HM contents in the surface sediments than the background values. The HM contamination in the sediments originated from human activities, primarily from industrial enterprises and with a large contribution from both agricultural and domestic sources. The most polluted HM with the highest ecological danger was Cd, followed by Cu, Zn, Ni, and As when the results of the four pollution evaluation indicators were combined. Consequently, the risk of contamination by HMs in inner-city river sediments should receive more attention.

Semantic community query in a large‐scale attributed graph based on an attribute cohesiveness optimization strategy

AbstractThe task of a semantic community query is to obtain a subgraph based on a given query vertex (or vertex set) and other query parameters in an attributed graph such that belongs to , contains and satisfies a predefined community cohesiveness model. In most cases, existing community query models based on the network structure for traditional attributed networks usually lack community semantics. However, the features of vertex attributes, especially the attributes of the query vertices, which are closely related to the community semantics, are rarely considered in an attributed graph. Existing community query algorithms based on both structure cohesiveness and attribute cohesiveness usually do not take the attributes of the query vertex as an important factor of the community cohesiveness model, which leads to weak semantics of the communities. This paper proposes a semantic community query method named in a large‐scale attributed graph. First, the k‐core structure model is adopted as the structure cohesiveness of our community query model to obtain a subgraph of the original graph. Second, we define attribute cohesiveness based on the average distance between the query vertices and other vertices in terms of attributes in the community to prune the subgraph and obtain the semantic community. In order to improve the community query efficiency in large‐scale attributed graphs, applies two heuristic pruning strategies. The experimental results show that our method outperforms the existing community query methods in multiple evaluation metrics and is ideal for querying semantic communities in large‐scale attributed graphs.

Multi-objective optimization of heat transfer performance of H-type fin with vortex generator based on NSGA-II

To further enhance the performance of heat exchangers and bolster waste heat recovery, vortex generators (VGs) of various shapes are added to the H-type fins. This study explores the impact of VGs’ attack angles and configurations on the thermal performance of H-type finned tubes under different Reynolds (Re) numbers. Additionally, multiple dimensionless evaluation metrics are introduced to assess the flow and heat transfer characteristics with VGs, and correlations for predicting the Nusselt (Nu) number and pressure drop at varying attack angles are developed. The results reveal that at low Re numbers, trapezoidal winglets yield the highest comprehensive performance factor (R-factor), while at high Re numbers, rectangular winglets prevail. Compared to trapezoidal winglet configurations, rectangular winglets enhance the Nu by 6.49%-6.67%. Among the various attack angle configurations, the 45° rectangular winglets exhibit the most favorable overall heat transfer performance, with a Nu increase of 12.07% compared to standard H-type fins. It is observed that installing more than two VGs can result in an R-value less than 1, suggesting that excessive VG placement in H-type fin heat exchangers is not advisable. Finally, the Non-dominated Sorting Genetic Algorithm II (NSGA-II) is employed to identify Pareto optimal points, which exhibit lower resistance and enhanced convective heat transfer. The Pareto optimal set, in comparison to different attack angle CFD data, shows a 31.27% reduction in resistance and a 15.72% increase in Nu.

Data-driven rapid detection of Helicobacter pylori infection through machine learning with limited laboratory parameters in Chinese primary clinics

BackgroundHelicobacter pylori (H. pylori) is a significant global health concern, posing a high risk for gastric cancer. Conventional diagnostic and screening approaches are inaccessible, invasive, inaccurate, time-consuming, and expensive in primary clinics. ObjectiveThis study aims to apply machine learning (ML) models to detect H. pylori infection using limited laboratory parameters from routine blood tests and to investigate the association of these biomarkers with clinical outcomes in primary clinics. MethodsA retrospective analysis with three ML and five ensemble models was conducted on 1409 adults from Hubei Provincial Hospital of Traditional Chinese Medicine. evaluating twenty-three blood test parameters and using the C14 urea breath test as the gold standard for diagnosing H. pylori infection. ResultsIn our comparative study employing three different feature selection strategies, Random Forest (RF) model exhibited superior performance over other ML and ensemble models. Multiple evaluation metrics underscored the optimal performance of the RF model (ROC = 0.951, sensitivity = 0.882, specificity = 0.906, F1 = 0.906, accuracy = 0.894, PPV = 0.908, NPV = 0.880) without feature selection. Key biomarkers identified through importance ranking and shapley additive Explanations (SHAP) analysis using the RF model without feature selection include White Blood Cell Count (WBC), Mean Platelet Volume (MPV), Hemoglobin (Hb), Red Blood Cell Count (RBC), Platelet Crit (PCT), and Platelet Count (PLC). These biomarkers were found to be significantly associated with the presence of H. pylori infection, reflecting the immune response and inflammation levels. ConclusionAbnormalities in key biomarkers could prompt clinical workers to consider H. pylori infection. The RF model effectively identifies H. pylori infection using routine blood tests, offering potential for clinical application in primary clinics. This ML approach can enhance diagnosis and screening, reducing medical burdens and reliance on invasive diagnostics.

RL-Chord: CLSTM-Based Melody Harmonization Using Deep Reinforcement Learning.

Automatic music generation is the combination of artificial intelligence and art, in which melody harmonization is a significant and challenging task. However, previous recurrent neural network (RNN)-based work fails to maintain long-term dependency and neglects the guidance of music theory. In this article, we first devise a universal chord representation with a fixed small dimension, which can cover most existing chords and is easy to expand. Then a novel melody harmonization system based on reinforcement learning (RL), RL-Chord, is proposed to generate high-quality chord progressions. Specifically, a melody conditional LSTM (CLSTM) model is put forward that learns the transition and duration of chords well, based on which RL algorithms with three well-designed reward modules are combined to construct RL-Chord. We compare three widely used RL algorithms (i.e., policy gradient, Q -learning, and actor-critic algorithms) on the melody harmonization task for the first time and prove the superiority of deep Q -network (DQN). Furthermore, a style classifier is devised to fine-tune the pretrained DQN-Chord for zero-shot Chinese folk (CF) melody harmonization. Experimental results demonstrate that the proposed model can generate harmonious and fluent chord progressions for diverse melodies. Quantitatively, DQN-Chord achieves better performance than the compared methods on multiple evaluation metrics, such as chord histogram similarity (CHS), chord tonal distance (CTD), and melody-chord tonal distance (MCTD).

Multi-Parameter Prediction of Solar Greenhouse Environment Based on Multi-Source Data Fusion and Deep Learning

In the process of agricultural production in solar greenhouses, the key to the healthy growth of greenhouse crops lies in accurately predicting environmental conditions. However, there are complex couplings and nonlinear relationships among greenhouse environmental parameters. This study independently developed a greenhouse environmental acquisition system to achieve a comprehensive method for the monitoring of the greenhouse environment. Additionally, it proposed a multi-parameter and multi-node environmental prediction model for solar greenhouses based on the Golden Jackal Optimization-Convolutional Neural Network-Bidirectional Gated Recurrent Unit-Self-Attention Mechanism (GCBS). The GCBS model successfully captures the complex nonlinear relationships in the greenhouse environment and accurately predicts changes in carbon dioxide concentration, air temperature and humidity, and soil temperature at different location nodes. To validate the performance of this model, we employed multiple evaluation metrics and conducted a comparative analysis with four baseline models. The results indicate that, while the GCBS model exhibits slightly higher computational time compared to the traditional Long Short-Term Memory (LSTM) network for time series prediction, it significantly outperforms the LSTM in terms of prediction accuracy for four key parameters, achieving improvements of 76.89%, 69.37%, 59.83%, and 56.72%, respectively, as measured by the Mean Absolute Error (MAE) metric.

Three-stage polyp segmentation network based on reverse attention feature purification with Pyramid Vision Transformer

Colorectal polyps serve as potential precursors of colorectal cancer and automating polyp segmentation aids physicians in accurately identifying potential polyp regions, thereby reducing misdiagnoses and missed diagnoses. However, existing models often fall short in accurately segmenting polyps due to the high degree of similarity between polyp regions and surrounding tissue in terms of color, texture, and shape. To address this challenge, this study proposes a novel three-stage polyp segmentation network, named Reverse Attention Feature Purification with Pyramid Vision Transformer (RAFPNet), which adopts an iterative feedback UNet architecture to refine polyp saliency maps for precise segmentation. Initially, a Multi-Scale Feature Aggregation (MSFA) module is introduced to generate preliminary polyp saliency maps. Subsequently, a Reverse Attention Feature Purification (RAFP) module is devised to effectively suppress low-level surrounding tissue features while enhancing high-level semantic polyp information based on the preliminary saliency maps. Finally, the UNet architecture is leveraged to further refine the feature maps in a coarse-to-fine approach. Extensive experiments conducted on five widely used polyp segmentation datasets and three video polyp segmentation datasets demonstrate the superior performance of RAFPNet over state-of-the-art models across multiple evaluation metrics.

DeepPGD: A Deep Learning Model for DNA Methylation Prediction Using Temporal Convolution, BiLSTM, and Attention Mechanism.

As part of the field of DNA methylation identification, this study tackles the challenge of enhancing recognition performance by introducing a specialized deep learning framework called DeepPGD. DNA methylation, a crucial biological modification, plays a vital role in gene expression analyses, cellular differentiation, and the study of disease progression. However, accurately and efficiently identifying DNA methylation sites remains a pivotal concern in the field of bioinformatics. The issue addressed in this paper is the presence of methylation in DNA, which is a binary classification problem. To address this, our research aimed to develop a deep learning algorithm capable of more precisely identifying these sites. The DeepPGD framework combined a dual residual structure involving Temporal convolutional networks (TCNs) and bidirectional long short-term memory (BiLSTM) networks to effectively extract intricate DNA structural and sequence features. Additionally, to meet the practical requirements of DNA methylation identification, extensive experiments were conducted across a variety of biological species. The experimental results highlighted DeepPGD's exceptional performance across multiple evaluation metrics, including accuracy, Matthews' correlation coefficient (MCC), and the area under the curve (AUC). In comparison to other algorithms in the same domain, DeepPGD demonstrated superior classification and predictive capabilities across various biological species datasets. This significant advancement in algorithmic prowess not only offers substantial technical support, but also holds potential for research and practical implementation within the DNA methylation identification domain. Moreover, the DeepPGD framework shows potential for application in genomics research, biomedicine, and disease diagnostics, among other fields.

MFTM-Informer: A multi-step prediction model based on multivariate fuzzy trend matching and Informer

Multi-step forecasting is a critical process in various fields, such as disaster warning and financial analysis. Nevertheless, achieving precise multi-step forecasting is challenging due to the intricate nature of the factors influencing the time series, most of which are highly nonlinear and nonstationary. In this paper, a multi-step forecasting model named MFTM-Informer, employing a multiple input multiple output strategy for multivariate trend matching is proposed. The dependent variable and the influencing factors are initially decomposed using multivariate variational modal decomposition to minimize noise. Afterwards, the decomposed data are reconstructed into multivariate trends and fluctuations using sample entropy, enabling the development of tailored forecasting strategies based on data characteristics. A multivariate trend is predicted using an enhanced pattern matching model, while the high-frequency fluctuation is modelled using Informer. Finally, the outcomes are combined to generate multi-step predictions. To validate the performance of the proposed model, we observed its performance on three real-world datasets, including Brent crude oil prices, European Union Allowance future prices, and Standard & Poor’s 500 index. Results indicate that the model surpasses all the benchmark models in terms of multiple evaluation metrics and forecast ranges, highlighting its effectiveness and robustness in multi-step forecasting.

FREE: Enhanced Feature Representation for Isotopic Envelope Evaluation in Top-Down Mass Spectra Deconvolution.

The aim of deconvolution of top-down mass spectra is to recognize monoisotopic peaks from the experimental envelopes in raw mass spectra. So accurate assessment of similarity between theoretical and experimental envelopes is a critical step in mass spectra data deconvolution. Existing evaluation methods primarily rely on intensity differences and m/z similarity, potentially lacking a comprehensive assessment. To overcome this constraint and facilitate a comprehensive and refined assessment of the similarity between theoretical and experimental envelopes, there exists an imperative to systematically explore and identify increasingly efficacious features for assessing this correspondence. We present enhanced feature representation for isotopic envelope evaluation (FREE) that derives diverse feature representations, encapsulating fundamental physical attributes of envelopes, including peak intensity and envelope shape. We trained FREE and evaluated its performance on both the ovarian tumor (OT) (human OT cells) data set and zebrafish (ZF) (brain in mature female ZF) data set. Specifically, comparing the state-of-art method, FREE demonstrates higher performance in multiple evaluation metrics across both the OT and ZF data sets, with a particular emphasis on precision, and it demonstrates accurate predictions of a greater number of positive envelopes among the top-ranked envelopes based on their scores. Moreover, within a cross-species data set of ZF, FREE identified a higher number of proteoform-spectrum matches (PrSMs), increasing the count from 50,795 to 52,927 compared to EnvCNN, the amalgamation of FREE with TopFD also exhibits a commendable capacity to discern 117,883 fragment ions, thus surpassing the 97,554 fragment ions identified through the application of EnvCNN in conjunction with TopFD. To further validate the performance of FREE, we have tested 10 a cross-species top-down proteomes containing 36 subdata set from ProteomeXchange. The results reveal that, after deconvolution with TopFD + FREE, TopPIC identifies more PrSMs across these 10 data sets in both the first and second rounds of experiments. These findings underscore the robustness and generalization capabilities of the FREE approach in diverse proteomes.

Predictive modeling of perioperative patient deterioration: combining unanticipated ICU admissions and mortality for improved risk prediction

ObjectiveThis paper presents a comprehensive analysis of perioperative patient deterioration by developing predictive models that evaluate unanticipated ICU admissions and in-hospital mortality both as distinct and combined outcomes.Materials and MethodsWith less than 1% of cases resulting in at least one of these outcomes, we investigated 98 features to identify their role in predicting patient deterioration, using univariate analyses. Additionally, multivariate analyses were performed by employing logistic regression (LR) with LASSO regularization. We also assessed classification models, including non-linear classifiers like Support Vector Machines, Random Forest, and XGBoost.ResultsDuring evaluation, careful attention was paid to the data imbalance therefore multiple evaluation metrics were used, which are less sensitive to imbalance. These metrics included the area under the receiver operating characteristics, precision-recall and kappa curves, and the precision, sensitivity, kappa, and F1-score. Combining unanticipated ICU admissions and mortality into a single outcome improved predictive performance overall. However, this led to reduced accuracy in predicting individual forms of deterioration, with LR showing the best performance for the combined prediction.DiscussionThe study underscores the significance of specific perioperative features in predicting patient deterioration, especially revealed by univariate analysis. Importantly, interpretable models like logistic regression outperformed complex classifiers, suggesting their practicality. Especially, when combined in an ensemble model for predicting multiple forms of deterioration. These findings were mostly limited by the large imbalance in data as post-operative deterioration is a rare occurrence. Future research should therefore focus on capturing more deterioration events and possibly extending validation to multi-center studies.ConclusionsThis work demonstrates the potential for accurate prediction of perioperative patient deterioration, highlighting the importance of several perioperative features and the practicality of interpretable models like logistic regression, and ensemble models for the prediction of several outcome types. In future clinical practice these data-driven prediction models might form the basis for post-operative risk stratification by providing an evidence-based assessment of risk.

Enhanced classification of microplastic polymers (polyethylene, polystyrene, low‐density polyethylene, polyhydroxyalkanoate) in waterbodies

AbstractThe contamination of microplastics (MPs) creates a substantial risk to both the environment and human health, necessitating the development of efficient methods for detecting and categorizing these micro pollutant particles. As a solution, Dense‐UNet with Convolutional Vision Transformer (Dense‐UNet‐CvT), a novel deep learning (DL)‐based model is proposed to detect and classify the MPs by performing the computer vision tasks. The main objective of this work is to enhance the detection accuracy in detecting the MPs classified from the input images. Initially, a holographic MPs image dataset comprising primary classes such as polyethylene (PE), polystyrene (PS), low‐density polyethylene (LDPE), polyhydroxyalkanoate (PHA) is collected for training and evaluating the research model. The images from the dataset are preprocessed by performing image resizing, Recursive Exposure based Sub‐Image Histogram Equalization (RESIHE)‐based image enhancement, Gaussian Adaptive Bilateral Filtering (GABF)‐based denoising to improve the visual quality of the images. The preprocessed images are applied for segmentation using the Dense‐UNet model for performing semantic segmentation. The CvT model is implemented to extract useful features and to perform classification on detecting the known and unknown classes of MPs labeled in the collected dataset. The MPs detection and classification performances are computed in terms of detection rate, accuracy, f1‐score, and precision. The Dense‐UNet‐CvT model achieved 98.22% detection rate, 98.59% accuracy, 98.35% f1‐score, and 98.76% precision. These performances are compared with the current models for proper validation, in which the research model outperformed all the compared models in terms of performance. Overall, the proposed Dense‐UNet‐CvT model demonstrates superior performance across multiple evaluation metrics, suggesting its effectiveness in detecting and classifying MPs contamination in holographic images.

Optimal Deep Neural Network Modified by Advanced Seasons Optimizer for Classification of Olympic Sports Images

Sports image classification has gained increasing significance in various domains, including athlete training, event management, and broadcasting. Nonetheless, accurately classifying these images remains a formidable challenge due to factors such as variations in lighting conditions, pose, and attire. This paper proposes an advanced deep learning approach to address this issue. The proposed method uses a modified Inception-V4 (IV4) architecture, which has exhibited remarkable success in image classification tasks. To further enhance the performance of the IV4 model, an advanced version of the seasons optimizer (ASO), a potent optimization algorithm, has been introduced, which adjusts the learning rate during training. This modified optimizer enables the IV4 model to converge more rapidly and achieve higher accuracy. An evaluation of the proposed method has been conducted using a comprehensive dataset comprising Olympic sports images. The experimental results demonstrate significant enhancements over existing approaches, attaining superior performance across multiple evaluation metrics. These findings underscore the potential of combining cutting-edge deep neural networks with advanced optimization techniques for the precise and efficient classification of sports images.

Crop Recommendation and Crop Yield Estimation Using Machine Learning and IoT

In the contemporary landscape of agriculture, the amalgamation of cutting-edge technologies such as machine learning (ML) and Internet of Things (IoT) has emerged as a transformative force, revolutionizing traditional farming practices. This paper introduces a comprehensive framework for crop recommendation and yield estimation, leveraging the power of ML algorithms and IoT devices to optimize agricultural operations. The dataset utilized in this study was meticulously curated from reputable sources including Kaggle and Google, encompassing a vast array of agricultural parameters ranging from soil characteristics to climate conditions. This rich dataset serves as the foundation for training and validating ML models, enabling robust analysis and prediction. A diverse range of ML algorithms was employed to process and interpret the dataset, with the objective of identifying the most proficient algorithm for crop recommendation and yield estimation tasks. Through rigorous experimentation and comparative analysis, the algorithm exhibiting superior performance across multiple evaluation metrics was meticulously chosen. Moreover, to bridge the gap between sophisticated ML techniques and practical application in the field, a user-friendly interface was developed using Flask, a micro web framework for Python. This intuitive interface empowers stakeholders to seamlessly access the trained model, facilitating real-time crop recommendations and yield estimations based on input parameters. The culmination of this research endeavor presents a paradigm shift in agricultural management, offering stakeholders unprecedented insights into crop selection and productivity optimization. By harnessing the potential of ML and IoT technologies, this framework not only enhances decision-making processes but also contributes to sustainable agricultural practices and global food security.

Multi-Class Network Anomaly Detection Using Machine Learning Techniques

Computer networks rely on Intrusion Detection Systems (IDSs) and Intrusion Prevention Systems (IPSs) to ensure the security, reliability, and availability of an organization. In recent years, various approaches were developed and implemented to create effective IDSs and IPSs. This paper specifically focuses on IDSs that utilize Machine Learning (ML) techniques for improved accuracy. ML-based IDSs have verified to be successful in discovering network attacks. However, their performance tends to decline when dealing with high-dimensional data spaces. It is essential to develop a suitable feature extraction strategy that could identify and remove irrelevant features that do not significantly classification process to address this issue. Additionally, many ML-based IDSs exhibit high false positive rates and poor detection accuracy when trained on unbalanced datasets. In this study, we analyze the UNSW-NB15 IDS, which will serve as the training and testing data for our models. In order to reduce the feature space and improve the efficiency of our analysis, we leverage a filter-based feature reduction method utilizing the Pearson correlation coefficient algorithm. By identifying and selecting only the most relevant features, we are able to streamline our dataset and focus on the variables that have the highest impact on our analysis. This approach not only reduces computational complexity but also improves the interpretability of our results by eliminating unnecessary noise from the data. After applying the feature reduction technique, we proceed to implement a range of machine learning methods to perform our classification task. These include well-known algorithms such as Stacking, Extra Trees, Multi-Layer Perceptron, XGBoost, K-Nearest Neighbors, Logistic Regression, Naïve Bayes, Support Vector Machine, Random Forest, and Decision Tree. By employing a diverse set of algorithms, we are able to explore different modeling approaches and evaluate their effectiveness in accurately classifying the various types of assaults. In order to assess the performance of our classification models, we utilize a range of specialized evaluation metrics such as Root Mean Square Error (RMSE), Mean Absolute Error (MAE), R2-Score, Mean Squared Error (MSE), Precision, F1-Score, Recall, and Accuracy. These metrics provide us with a comprehensive understanding of how well our models are performing across different dimensions, including the accuracy of predictions, the level of precision in classifying different assault types, and the overall goodness-of-fit of our models. By considering multiple evaluation metrics, we are able to gain a more nuanced understanding of the strengths and weaknesses of each algorithm and make informed decisions about their suitability for our classification task. These metrics deliver a complete evaluation of the classifiers’ effectiveness in detecting community intrusions.

RAGE-Net: Enhanced retinal vessel segmentation U-shaped network using Gabor convolution

Extracting vessel morphology from fundus images is pivotal in acquiring pathological insights and enabling early diagnosis of retinal disorders. Manual segmentation of retinal vessels requires a high degree of expertise and is notably time-intensive. Although existing deep learning techniques for retinal vessel segmentation predominantly hinge on U-shaped convolutional neural networks, significant headway has been made, complexities persist in delineating faint, low-contrast vessels amidst noisy backgrounds. To confront these challenges, we propose an innovative U-shaped convolutional neural network fortified with oriented priors, labeled as the Receptive Field Aggregating Gabor Enhance Network (RAGE-Net). We revamp the conventional U-shaped convolutional network with a foundation in Gabor wavelet and Gabor convolutional network, introducing a Gabor Matching Enhance Architecture (GMEA) amalgamated into the U-shaped convolutional network. This architecture comprises two distinct modules. Initially, a Dual-scale Gabor Enhance Module (DGEB) is introduced to bolster vessel continuity and effectively fortify delicate vessels by integrating oriented feature enhancement through Gabor convolution. Subsequently, a Receptive Field Pyramid Module (RPM) is proposed to supplant the escalation of scale count in the Gabor filter bank for vessel alignment, also serving as feature fusion to enhance the network's comprehensive vessel discernment. In comparison to U-Net, our model boasts fewer parameters and surpasses in terms of sensitivity, accuracy, and F1 score on the DRIVE dataset by 3.58%, 0.34%, and 2.29%, respectively. Our model showcases stellar performance across three public datasets: DRIVE, STARE, and CHASE_DB1, with sensitivity values of 0.8172, 0.8126, and 0.8540 and accuracy figures of 0.9708, 0.9725, and 0.9757, respectively. The analysis on three datasets reveals that our model demonstrates distinct strengths in AUC and Se as a result of integrating GMEA, which comprises RPM and DGEB, into U-shaped networks. However, it does not reduce overall accuracy to improve the model's ability to perceive weak contrast and small blood vessels. While maintaining superior Acc, our model also demonstrates advancements in Sp and F1 score, indicating a balanced progression in multiple evaluation metrics.

Application of deep learning and statistical methods in predicting Taiwanese stock trends

Since its inception, the stock market has been a topic of considerable interest. Its variation and the complexity of integrating technology into the stock market have made it difficult for stock market trends to be fully understood. Various metrics and analytical approaches have been proposed in response to such changes, ranging from purely technical metrics to hardware upgrades. The widespread application of deep learning in the stock market, from basic metrics (opening price, closing price, highest price, lowest price, trading volume) to machine learning in sentiment analysis, further increases the possibility of increasing profits. Some front-end techniques, such as noise reduction through mathematical models, enhance the accuracy of deep learning models. However, few studies have centered on predicting long-term stock price changes. The traditional moving average (MA) cannot rapidly reflect drastic changes on its curve even though it can display trends; therefore, this study proposes an MA-based approach that improves the 200-day MA such that its delayed response to actual prices in real-time can be overcome. This deep learning model training was performed by combining 200-day MA data with two other types of MA data, thereby creating a new approach to metric analysis. The sample consisted of stocks of 13 Taiwanese companies with a high market cap: Taiwan Semiconductor Manufacturing Co., Ltd., MediaTek Inc., Chunghwa Telecom Co., Ltd., Fubon Financial Holding Co., Ltd., Cathay Financial Holding Co., Ltd., Nan Ya Plastics Corp., United Microelectronics Corp., Delta Electronics, Inc., CTBC Financial Holding Co., Ltd., Mega Financial Holding Co., Ltd., Formosa Chemicals &amp; Fibre Corp., Hon Hai Precision Industry Co., Ltd., and Formosa Plastics Corp. Through multiple evaluation metrics, the experimental results revealed that the proposed model performed better in general than the traditional MA model for all stocks.

Predicting the Potential Distribution of Quercus oxyphylla in China under Climate Change Scenarios

Global climate changes are expected to profoundly shape species distribution. Quercus oxyphylla, a valuable evergreen broad-leaved tree species, is rigorously conserved and managed in China owing to its substantial scientific, economic, and ecological value. However, the impact of projected climate change on its future distribution and potential climatic drivers remains unclear. Here, a maximum entropy model (MaxEnt) was used to explore the distribution of Q. oxyphylla in China under current conditions and three future scenarios (SSP1-2.6, SSP2-4.5, and SSP5-8.5) for the 2050s and 2070s. We optimized the model using the ‘ENMeval’ package to obtain the best parameter combination (RM = 1, FC = LQHPT), and multiple evaluation metrics (AUC ≥ 0.9; TSS ≥ 0.6; Kappa ≥ 0.75) verified the high accuracy of the model and the reliability of the prediction results. We found the following: (1) The potential distribution of Q. oxyphylla spans across 28 provinces in China under current climatic conditions, predominantly in southern regions, with Sichuan exhibiting the largest suitable area for survival. The total suitable habitat covers 244.98 × 104 km2, comprising highly, moderately, and poorly suitable habitats of 51.66 × 104 km2, 65.98 × 104 km2, and 127.34 × 104 km2, respectively. (2) Under future climate conditions, the overall geographical boundaries of Q. oxyphylla are predicted to remain similar to the present one, with an increase of 10.29% in the 2050s and 11.31% in the 2070s. In the 2050s, the total suitable habitats for Q. oxyphylla under the three scenarios (SSP1-2.6, SSP2-4.5, and SSP5-8.5) might increase by 8.83%, 9.62%, and 12.42%, while in the 2070s they might increase by 10.39%, 17.21%, and 6.33%, respectively. (3) Moreover, the centroid of the suitable area is expected to migrate southwestward under the three scenarios in the future. (4) Annual precipitation, isothermality, and temperature annual range emerged as the main factors influencing the distribution of Q. oxyphylla, with contributions of 55.9%, 25.7%, and 13.5%, respectively. Our findings refined the spatial arrangement of Q. oxyphylla growth and revealed its climate resilience. This suggested that under climate change, Sichuan and Shaanxi are the optimal regions for cultivation and management, while appropriate conservation strategies should be formulated in Tibet and Hubei.