Highest F1 Score Research Articles

Formula: see text]-[Formula: see text]: A Nonparametric Model for Neural Power Spectra Decomposition.

The power spectra estimated from the brain recordings are the mixed representation of aperiodic transient activity and periodic oscillations, i.e., aperiodic component (AC) and periodic component (PC). Quantitative neurophysiology requires precise decomposition preceding parameterizing each component. However, the shape, statistical distribution, scale, and mixing mechanism of AC and PCs are unclear, challenging the effectiveness of current popular parametric models such as FOOOF, IRASA, BOSC, etc. Here, ξ- π was proposed to decompose the neural spectra by embedding the nonparametric spectra estimation with penalized Whittle likelihood and the shape language modeling into the expectation maximization framework. ξ- π was validated on the synthesized spectra with loss statistics and on the sleep EEG and the large sample iEEG with evaluation metrics and neurophysiological evidence. Compared to FOOOF, both the simulation presenting shape irregularities and the batch simulation with multiple isolated peaks indicated that ξ- π improved the fit of AC and PCs with less loss and higher F1-score in recognizing the centering frequencies and the number of peaks; the sleep EEG revealed that ξ- π produced more distinguishable AC exponents and improved the sleep state classification accuracy; the iEEG showed that ξ- π approached the clinical findings in peak discovery. Overall, ξ- π offered good performance in the spectra decomposition, which allows flexible parameterization using descriptive statistics or kernel functions. ξ- π is a seminal tool for brain signal decoding in fields such as cognitive neuroscience, brain-computer interface, neurofeedback, and brain diseases.

A comparative analysis of cervical cancer diagnosis using machine learning techniques

This study undertakes a comprehensive analysis of cervical cancer diagnosis using machine learning (ML) techniques. We start by introducing the critical importance of early and accurate diagnosis of cervical cancer, a significant health issue globally. The objective of this research is to compare the effectiveness of three ML algorithms: K-nearest neighbors (KNN), linear support vector machine (SVM), and Naive Bayes classifier, in predicting biopsy results for cervical cancer. Our methodology involves utilizing a substantial dataset to train and test these algorithms, focusing on performance measures like accuracy, precision, recall, F1 score, and the area under the receiver operating characteristic curve (AUC). The findings reveal that KNN demonstrates superior performance, with high precision, recall, accuracy, and F1 score, alongside a notable AUC. This suggests KNN's potential utility in clinical applications for cervical cancer prognosis. Meanwhile, linear SVM and Naive Bayes exhibit certain limitations, indicating a need for further optimization. This study highlights the promising role of ML in enhancing medical diagnostic processes, particularly in oncology.

Whose work matters? A tool for identifying and developing more inclusive physics textbooks

The lack of representational diversity and role models in physics, including in our textbooks and curricular materials, is an oft-cited contributing factor to the continuing dramatic under-representation of women and people of color in physics. In this work, we develop an automated, Python-based tool for identifying the names and demographics of scientists who are mentioned in indices and chapters of physics textbooks, enabling authors, publishers, and users of physics textbooks to rapidly analyze the demographics of these texts. We quantitatively validate the automated tool using standard machine learning metrics, attaining high accuracy, precision, recall, and F1 scores. The tool is then used to demonstrate two of the many potential applications: examining whose work is mentioned in the entire collection of textbooks used in a representative four-year undergraduate physics major curriculum as well as an analysis of the demographics of scientists mentioned in a selection of ten introductory physics textbooks. Both of the sample analyses result in a similar portrait, showing that the undergraduate physics textbooks examined in this work focus overwhelmingly on work attributed to White men of European, British, and North American descent. This work points to an urgent need for the physics education community, including textbook publishers, authors, and adopters, to work together to broaden our portrayals of physics to reflect the vast diversity of scientists, both historically and contemporaneously, who are working in this field.

Segmentation of liver CT images based on weighted medical transformer model.

Deep convolutional neural networks have made significant strides in the field of medical image segmentation. Although existing convolutional structures enhance performance by leveraging local image information, they often lose the interdependence information between contexts. Therefore, the article utilizes the multi-attention mechanism of the Transformer structure to more comprehensively express relationships between contexts and introduced the Transformer network architecture into the field of medical image segmentation. Most models based on this Transformer structure typically require large datasets for training. However, in the medical field, the limited size of datasets makes training models with the Transformer structure challenging. To address this, the article propose a Weighted Medical Transformer (WMT) model that imposes low requirements on dataset quantity. The weighting mechanism in the WMT model aims to improve the issue of inaccurate relative positional coding when dealing with small medical datasets. Additionally, a coarse-grained and fine-grained segmentation mechanism is introduced, focusing on both the detailed aspects within image blocks and the boundary information connecting blocks. Experimental results on a liver dataset demonstrate that the model achieves F1 and IoU scores of 88.48% and 79.41%, respectively. Results on the MoNuSeg dataset show comparable high F1 and IoU scores of 79.58% and 66.19%, respectively. The model's accuracy surpasses that of U-Net++ and U-Net models. Compared to other models, this approach is applicable to scenarios with limited datasets, exhibiting high execution efficiency and accuracy.

Empowering Smart City IoT Network Intrusion Detection with Advanced Ensemble Learning-based Feature Selection

This study presents an advanced methodology tailored for enhancing the performance of Intrusion Detection Systems (IDS) deployed in Internet of Things (IoT) networks within smart city environments. Through the integration of advanced techniques in data preprocessing, feature selection, and ensemble classification, the proposed approach addresses the unique challenges associated with securing IoT networks in urban settings. Leveraging techniques such as SelectKBest, Recursive Feature Elimination (RFE), and Principal Component Analysis (PCA), combined with the Gradient-Based One Side Sampling (GOSS) technique for model training, the methodology achieves high accuracy, precision, recall, and F1 score across various evaluation scenarios. Evaluation on the UNSW-NB15 dataset demonstrates the effectiveness of the proposed approach, with comparative analysis showcasing its superiority over existing techniques.

Forecasting of Flood in the Non-Tidal River of Northern Regions of Bangladesh Using Machine Learning-Based Approach

Forecasting of Flood in the Non-Tidal River of Northern Regions of Bangladesh Using Machine Learning-Based Approach

"FLOODGUARD: An IoT-Powered Real-Time Flash Flood Monitoring and Forecasting System"

Abstract: In recent times, the surge in flooding incidents has become a pressing global issue, inflicting extensive damage on both lives and livelihoods. While floods remain an inevitable force of nature, their devastating impact can be alleviated. With the aid of cutting-edge technologies like the Internet of Things (IoT), proactive flood forecasting has become increasingly feasible. Through the seamless integration of IoT data, populations can receive early warnings and evacuate to safety, safeguarding both lives and valuables. A real-time solution is imperative, leveraging IoT data streams to deliver timely flood alerts. Our study presents an IoT-driven prototype designed to collect hydrological data from rivers, encompassing metrics such as water flow, level, and discharge, alongside meteorological parameters like temperature, humidity, wind speed, and direction. Employing a Long Short-Term Memory (LSTM) model, we analyze and classify collected data based on water discharge, level, rainfall, and temperature, predicting flood events as "no alert," "yellow alert," "orange alert," or "red alert." Given the dynamic nature of rivers, accurately measuring total water discharge poses challenges. To address this, we propose a novel methodology utilizing water flow, sectional area width, and average depth. Furthermore, we tackle the complexities associated with accurately measuring rainfall due to erratic weather conditions. Our system demonstrates robust performance in predicting flood event states, achieving high F1-scores of 97% for "no alert," 97% for "yellow alert," 96% for "orange alert," and 98% for "red alert.

Development of Athlete Health Monitoring and Early Warning System Based on Sensor Data

An athlete health monitoring and early warning system based on sensor data utilizes wearable sensors to track various physiological parameters, such as heart rate, body temperature, oxygen saturation, and movement patterns. These sensors continuously collect real-time data during training sessions and competitions. Machine learning algorithms analyze the data to detect deviations from normal patterns and identify potential indicators of injury, fatigue, or overtraining. By monitoring athletes' health parameters over time and comparing them to established baseline values, the system can provide early warnings of potential health issues or performance declines. Coaches and medical staff can then intervene promptly with appropriate interventions, adjustments to training loads, or rest periods to prevent injuries and optimize athletes' performance and well-being. . The paper introduces the Reliable Fuzzy Sensor Optimized Automated Monitoring (RFSO-DL) system, a novel approach to athlete health monitoring leveraging fuzzy logic and sensor data optimization techniques. RFSO-DL offers a comprehensive solution for real-time assessment of athlete well-being by analyzing physiological measurements and activity levels. Through a series of experiments, the system demonstrates high accuracy, precision, recall, and F1 score in classifying athlete health status. Results indicate RFSO-DL's effectiveness in capturing subtle variations in health status and providing timely interventions. Through experiments, the system achieves impressive results, with accuracy ranging from 90.5% to 94.1%, precision from 85.8% to 91.2%, recall from 91.7% to 95.6%, and F1 score from 88.6% to 93.3%. These findings underscore RFSO-DL's potential to revolutionize athlete health monitoring by accurately classifying health status based on physiological measurements and activity levels.

Automated abdominal CT contrast phase detection using an interpretable and open-source artificial intelligence algorithm.

To develop and validate an open-source artificial intelligence (AI) algorithm to accurately detect contrast phases in abdominal CT scans. Retrospective study aimed to develop an AI algorithm trained on 739 abdominal CT exams from 2016 to 2021, from 200 unique patients, covering 1545 axial series. We performed segmentation of five key anatomic structures-aorta, portal vein, inferior vena cava, renal parenchyma, and renal pelvis-using TotalSegmentator, a deep learning-based tool for multi-organ segmentation, and a rule-based approach to extract the renal pelvis. Radiomics features were extracted from the anatomical structures for use in a gradient-boosting classifier to identify four contrast phases: non-contrast, arterial, venous, and delayed. Internal and external validation was performed using the F1 score and other classification metrics, on the external dataset "VinDr-Multiphase CT". The training dataset consisted of 172 patients (mean age, 70 years ± 8, 22% women), and the internal test set included 28 patients (mean age, 68 years ± 8, 14% women). In internal validation, the classifier achieved an accuracy of 92.3%, with an average F1 score of 90.7%. During external validation, the algorithm maintained an accuracy of 90.1%, with an average F1 score of 82.6%. Shapley feature attribution analysis indicated that renal and vascular radiodensity values were the most important for phase classification. An open-source and interpretable AI algorithm accurately detects contrast phases in abdominal CT scans, with high accuracy and F1 scores in internal and external validation, confirming its generalization capability. Contrast phase detection in abdominal CT scans is a critical step for downstream AI applications, deploying algorithms in the clinical setting, and for quantifying imaging biomarkers, ultimately allowing for better diagnostics and increased access to diagnostic imaging. Digital Imaging and Communications in Medicine labels are inaccurate for determining the abdominal CT scan phase. AI provides great help in accurately discriminating the contrast phase. Accurate contrast phase determination aids downstream AI applications and biomarker quantification.

Design and Evaluation of English Vocabulary Learning Aids Based on Word Vector Modelling

English vocabulary learning aids based on word vector modeling involve creating tools that leverage advanced techniques to enhance vocabulary acquisition. Word vector modeling, often using methods like Word2Vec or GloVe, represents words as high-dimensional vectors capturing semantic relationships. These models can power vocabulary learning aids by offering context-based word suggestions, personalized word quizzes, or interactive visualizations of word associations.. The paper introduces the Hierarchical Spiking Vocabulary Deep Learning (HSV-DL) framework, a novel approach aimed at enhancing vocabulary learning and classification tasks. With word vector modeling and spiking neural networks, HSV-DL offers a sophisticated methodology for accurately categorizing vocabulary words into their respective semantic categories. The Hierarchical Spiking Vocabulary Deep Learning (HSV-DL) framework introduces novel methodologies for vocabulary learning and classification tasks, achieving outstanding performance metrics. Experimental results demonstrate high accuracy (95%), precision (96%), recall (94%), and F1-score (95%) in categorizing vocabulary words into their semantic categories. Moreover, HSV-DL exhibits robustness to noise and efficient resource utilization, showcasing its potential for real-world applications in natural language processing.

A Learning-Based Approach to Static Program Slicing

Traditional program slicing techniques are crucial for early bug detection and manual/automated debugging of online code snippets. Nevertheless, their inability to handle incomplete code hinders their real-world applicability in such scenarios. To overcome these challenges, we present NS-Slicer, a novel learning-based approach that predicts static program slices for both complete and partial code Our tool leverages a pre-trained language model to exploit its understanding of fine-grained variable-statement dependencies within source code. With this knowledge, given a variable at a specific location and a statement in a code snippet, NS-Slicer determines whether the statement belongs to the backward slice or forward slice, respectively. We conducted a series of experiments to evaluate NS-Slicer's performance. On complete code, it predicts the backward and forward slices with an F1-score of 97.41% and 95.82%, respectively, while achieving an overall F1-score of 96.77%. Notably, in 85.20% of the cases, the static program slices predicted by NS-Slicer exactly match entire slices from the oracle. For partial programs, it achieved an F1-score of 96.77%–97.49% for backward slicing, 92.14%–95.40% for forward slicing, and an overall F1-score of 94.66%–96.62%. Furthermore, we demonstrate NS-Slicer's utility in vulnerability detection (VD), integrating its predicted slices into an automated VD tool. In this setup, the tool detected vulnerabilities in Java code with a high F1-score of 73.38%. We also include the analyses studying NS-Slicer’s promising performance and limitations, providing insights into its understanding of intrinsic code properties such as variable aliasing, leading to better slicing.

Comparison and Optimal Method of Detecting the Number of Maize Seedlings Based on Deep Learning

Effective agricultural management in maize production operations starts with the early quantification of seedlings. Accurately determining plant presence allows growers to optimize planting density, allocate resources, and detect potential growth issues early on. This study provides a comprehensive analysis of the performance of various object detection models in maize production, with a focus on the effects of planting density, growth stages, and flight altitudes. The findings of this study demonstrate that one-stage models, particularly YOLOv8n and YOLOv5n, demonstrated superior performance with AP50 scores of 0.976 and 0.951, respectively, outperforming two-stage models in terms of resource efficiency and seedling quantification accuracy. YOLOv8n, along with Deformable DETR, Faster R-CNN, and YOLOv3-tiny, were identified for further examination based on their performance metrics and architectural features. The study also highlights the significant impact of plant density and growth stage on detection accuracy. Increased planting density and advanced growth stages (particularly V6) were associated with decreased model accuracy due to increased leaf overlap and image complexity. The V2–V3 growth stages were identified as the optimal periods for detection. Additionally, flight altitude negatively affected image resolution and detection accuracy, with higher altitudes leading to poorer performance. In field applications, YOLOv8n proved highly effective, maintaining robust performance across different agricultural settings and consistently achieving rRMSEs below 1.64% in high-yield fields. The model also demonstrated high reliability, with Recall, Precision, and F1 scores exceeding 99.00%, affirming its suitability for practical agricultural use. These findings suggest that UAV-based image collection systems employing models like YOLOv8n can significantly enhance the accuracy and efficiency of seedling detection in maize production. The research elucidates the critical factors that impact the accuracy of deep learning detection models in the context of corn seedling detection and selects a model suited for this specific task in practical agricultural production. These findings offer valuable insights into the application of object detection technology and lay a foundation for the future development of precision agriculture, particularly in optimizing deep learning models for varying environmental conditions that affect corn seedling detection.

A novel semi-supervised model for pre-impact fall detection with limited fall data

Due to the high prevalence and severe consequences of falls among older people, pre-impact fall detection algorithms which aim to detect falls before body-ground impacts are of critical importance. The success of existing algorithms based on wearable sensors relies heavily on numerous labeled data from both falls and activities of daily living (ADLs). However, fall data is much more difficult to acquire compared with ADLs, and moreover, labeling falling period apart from ADLs is laborious and time-consuming. To this end, we proposed a novel semi-supervised model for pre-impact fall detection, called Semi-PFD, which maximizes the utilization of easily obtainable ADL data and reduces the dependency on labeled fall data. This model was evaluated on two large-scale public fall datasets (KFall and SisFall) and compared with its supervised baseline under different ratios of fall data. Cross-validation results showed that Semi-PFD outperformed the supervised baseline across all conditions on both datasets. More importantly, Semi-PFD achieved considerable improvement of 2–4% F1-score when the ratios of fall data were low (≤23.1%). Remarkably, with only 76.9% of fall data, Semi-PFD achieved comparable and even higher F1-scores than the state-of-the-art benchmark models which utilized 100% of fall data. We further observed that the incorporation of unsupervised training in Semi-PFD mitigated the typical overconfidence problem associated with supervised training without affecting lead time. These findings underscore Semi-PFD's practical potential, reducing the burden of fall data collection and labeling while maintaining superior performance.

Non-Intrusive Load Monitoring Based on Multiscale Attention Mechanisms

With the development of smart grids and new power systems, the combination of non-intrusive load identification technology and smart home technology can provide users with the operating conditions of home appliances and equipment, thus reducing home energy loss and improving users’ ability to demand a response. This paper proposes a non-intrusive load decomposition model with a parallel multiscale attention mechanism (PMAM). The model can extract both local and global feature information and fuse it through a parallel multiscale network. This improves the attention mechanism’s ability to capture feature information over long time periods. To validate the model’s decomposition ability, we combined the PMAM model with four benchmark models: the Long Short-Term Memory (LSTM) recurrent neural network model, the Time Pooling-based Load Disaggregation Model (TPNILM), the Extreme Learning Machine (ELM), and the Load Disaggregation Model without Parallel Multi-scalar Attention Mechanisms (UNPMAM). The model was trained on the publicly available UK-DALE dataset and tested. The models’ test results were quantitatively evaluated using a confusion matrix. This involved calculating the F1 score of the load decomposition. A higher F1 score indicates better model decomposition performance. The results indicate that the PMAM model proposed in this paper maintains an F1 score above 0.9 for the decomposition of three types of electrical equipment under the same household user, which is 3% higher than that of the other benchmark models on average. In the cross-household test, the PMAM also demonstrated a better decomposition ability, with the F1 score maintained above 0.85, and the mean absolute error (MAE) decreased by 5.3% on average compared with that of the UNPMAM.

Machine learning approach for predicting cardiovascular disease in Bangladesh: evidence from a cross-sectional study in 2023

BackgroundCardiovascular disorders (CVDs) are the leading cause of death worldwide. Lower- and middle-income countries (LMICs), such as Bangladesh, are also affected by several types of CVDs, such as heart failure and stroke. The leading cause of death in Bangladesh has recently switched from severe infections and parasitic illnesses to CVDs.Materials and methodsThe study dataset comprised a random sample of 391 CVD patients' medical records collected between August 2022 and April 2023 using simple random sampling. Moreover, 260 data points were collected from individuals with no CVD problems for comparison purposes. Crosstabs and chi-square tests were used to determine the association between CVD and the explanatory variables. Logistic regression, Naïve Bayes classifier, Decision Tree, AdaBoost classifier, Random Forest, Bagging Tree, and Ensemble learning classifiers were used to predict CVD. The performance evaluations encompassed accuracy, sensitivity, specificity, and area under the receiver operator characteristic (AU-ROC) curve.ResultsRandom Forest had the highest precision among the five techniques considered. The precision rates for the mentioned classifiers are as follows: Logistic Regression (93.67%), Naïve Bayes (94.87%), Decision Tree (96.1%), AdaBoost (94.94%), Random Forest (96.15%), and Bagging Tree (94.87%). The Random Forest classifier maintains the highest balance between correct and incorrect predictions. With 98.04% accuracy, the Random Forest classifier achieved the best precision (96.15%), robust recall (100%), and high F1 score (97.7%). In contrast, the Logistic Regression model achieved the lowest accuracy of 95.42%. Remarkably, the Random Forest classifier achieved the highest AUC value (0.989).ConclusionThis research mainly focused on identifying factors that are critical in impacting patients with CVD and predicting CVD risk. It is strongly advised that the Random Forest technique be implemented in a system for predicting cardiac diseases. This research may change clinical practice by providing doctors with a new instrument to determine a patient’s CVD prognosis.

Joint transformer architecture in brain 3D MRI classification: its application in Alzheimer's disease classification.

Alzheimer's disease (AD), a neurodegenerative disease that mostly affects the elderly, slowly impairs memory, cognition, and daily tasks. AD has long been one of the most debilitating chronic neurological disorders, affecting mostly people over 65. In this study, we investigated the use of Vision Transformer (ViT) for Magnetic Resonance Image processing in the context of AD diagnosis. ViT was utilized to extract features from MRIs, map them to a feature sequence, perform sequence modeling to maintain interdependencies, and classify features using a time series transformer. The proposed model was evaluated using ADNI T1-weighted MRIs for binary and multiclass classification. Two data collections, Complete 1Yr 1.5T and Complete 3Yr 3T, from the ADNI database were used for training and testing. A random split approach was used, allocating 60% for training and 20% for testing and validation, resulting in sample sizes of (211, 70, 70) and (1378, 458, 458), respectively. The performance of our proposed model was compared to various deep learning models, including CNN with BiL-STM and ViT with Bi-LSTM. The suggested technique diagnoses AD with high accuracy (99.048% for binary and 99.014% for multiclass classification), precision, recall, and F-score. Our proposed method offers researchers an approach to more efficient early clinical diagnosis and interventions.

Performance Assessment of ChatGPT versus Bard in Detecting Alzheimer's Dementia.

Large language models (LLMs) find increasing applications in many fields. Here, three LLM chatbots (ChatGPT-3.5, ChatGPT-4, and Bard) are assessed in their current form, as publicly available, for their ability to recognize Alzheimer's dementia (AD) and Cognitively Normal (CN) individuals using textual input derived from spontaneous speech recordings. A zero-shot learning approach is used at two levels of independent queries, with the second query (chain-of-thought prompting) eliciting more detailed information than the first. Each LLM chatbot's performance is evaluated on the prediction generated in terms of accuracy, sensitivity, specificity, precision, and F1 score. LLM chatbots generated a three-class outcome ("AD", "CN", or "Unsure"). When positively identifying AD, Bard produced the highest true-positives (89% recall) and highest F1 score (71%), but tended to misidentify CN as AD, with high confidence (low "Unsure" rates); for positively identifying CN, GPT-4 resulted in the highest true-negatives at 56% and highest F1 score (62%), adopting a diplomatic stance (moderate "Unsure" rates). Overall, the three LLM chatbots can identify AD vs. CN, surpassing chance-levels, but do not currently satisfy the requirements for clinical application.

Feature Extraction from Regenerated EEG – A Better Approach for ICA Based Eye Blink Artifact Detection

Purpose: Independent Component Analysis (ICA) decomposition is a commonly used technique for eye blink artifact detection from Electroencephalogram (EEG) signals. Feature extraction from the decomposed ICs is a prime step for blink detection. This paper presents a new model of eye blink detection for ICA based approach, where the decomposed ICs are projected to their corresponding EEG segments (ReEEG), and feature extraction is performed on the ReEEG instead of the IC. ReEEG represents the eye blink activity more distinctly. Hence, ReEEG-based feature extraction is more potential in detecting eye blink artifacts than the traditional IC-based feature extraction.
 Materials and Methods: This paper employs twelve EEG features to substantiate the superiority of ReEEG over IC. Support Vector Machine (SVM) is used as a classifier. A dataset, having 2638 clinical EEG epochs, is employed. All the considered twelve features are extracted from ReEEG and fed to SVM one at a time for blink detection. Then the obtained results are compared with an IC-based model with the same features.
 Results: The comparison reveals the success of the proposed ReEEG-based blink detection approach over the traditional IC-based approach. Accuracy, precision, recall, and f1 scores are calculated as performance measuring metrics. For almost all features, ReEEG-based approach achieved up to 12.25% higher accuracy, 24.95% higher precision, 13.49% higher recall, and 12.89% higher f1 score than the IC-based traditional method.
 Conclusion: The proposed model will be useful for researchers in dealing with the eye blink artifacts of EEG signals with more efficacy.

Detection of Deepfake Media Using a Hybrid CNN–RNN Model and Particle Swarm Optimization (PSO) Algorithm

Deepfakes are digital audio, video, or images manipulated using machine learning algorithms. These manipulated media files can convincingly depict individuals doing or saying things they never actually did. Deepfakes pose significant risks to our lives, including national security, financial markets, and personal privacy. The ability to create convincing deep fakes can also harm individuals’ reputations and can be used to spread disinformation and fake news. As such, there is a growing need for reliable and accurate methods to detect deep fakes and prevent their harmful effects. In this paper, a hybrid convolutional neural network (CNN) and recurrent neural network (RNN) with a particle swarm optimization (PSO) algorithm is utilized to demonstrate a deep learning strategy for detecting deepfake videos. High accuracy, sensitivity, specificity, and F1 score were attained by the proposed approach when tested on two publicly available datasets: Celeb-DF and the Deepfake Detection Challenge Dataset (DFDC). Specifically, the proposed method achieved an average accuracy of 97.26% on Celeb-DF and an average accuracy of 94.2% on DFDC. The results were compared to other state-of-the-art methods and showed that the proposed method outperformed many. The proposed method can effectively detect deepfake videos, which is essential for identifying and preventing the spread of manipulated content online.

A multi-task learning risk assessment method for the chemical process industry

Dynamic risk assessment is essential to prevent major accidents in the chemical process industry. The Prediction of accident risk categories, risk likelihood, and risk severity are three fundamental tasks for risk assessment. Single-task learning-based approaches, which merely analyze accident occurrence or quantify the likelihood and severity of an accident risk, often tend to ignore task relevance and inhibit the sharing of critical feature information. This oversight might lead to poor prediction performance. In this study, we propose a multi-task learning model, i.e., Robustly Progressive Layered Extraction (RoPLE), to systematically predict accident risk category, risk likelihood, and risk severity. The RoPLE consists of two parts: a robustly optimized pre-trained module for extracting semantic features from hazard analysis reports (unstructured data) and a progressive layered extraction module for fusing semantic features with features of leading indicators (structured data), which portray the risk state of equipment, workers, environment, and management in the chemical process industry. The progressive layered extraction module contains a multi-level extraction network and a tower network for these three tasks. The multi-level extraction network extracts higher-level shared information, while the tower network extracts task-specific feature information. Empirical experiments demonstrate the feasibility of our model on a real risk assessment dataset in Beijing’s chemical process industry. The RoPLE obtains high F1-scores of 63.1%, 79.5%, and 86.2% for the three tasks, and significantly outperforms other state-of-the-art single-task methods.