Skewness Research Articles

Examining the impact of intellectual capital on underwriting risk of insurance companies in Pakistan: static and dynamic panel regression

Purpose This study aims to examine the impact of intellectual capital on the underwriting risk of insurance companies in Pakistan. Design/methodology/approach The study used a quantitative research approach and a longitudinal research design, gathering data from 23 insurance companies listed on the Pakistan stock exchange from 2010 to 2022. The value-added intellectual coefficient (VAIC) was used to measure intellectual capital (IC), and unbalanced panel data were analyzed using static and dynamic regression analyses. Findings The findings demonstrate a significant association between intellectual capital and underwriting risk in insurance companies in Pakistan. Specific components of intellectual capital, such as human capital efficiency (HCE), structural capital efficiency (SCE) and capital employed efficiency (CEE), have a strong negative impact on underwriting risk. Control variables like return on assets, insurer size and leverage also showed significant relationships with underwriting risk. Originality/value This research provides new insights into the theoretical understanding of the insurance industry by establishing a direct link between intellectual capital and underwriting risk in the context of Pakistan. It suggests that by improving aspects of intellectual capital, specifically HCE, SCE and CEE, policymakers and managers can reduce underwriting risk.

Genome‐Wide Association Mapping and Genomic Prediction of Septoria nodorum Blotch Resistance in Central European Winter Wheat (Triticum aestivum L.)

ABSTRACTSeptoria nodorum blotch (SNB) is a fungal disease of wheat caused by the necrotrophic fungus Parastagonospora nodorum (Berk.), considered as one of the most devastating fungal diseases affecting winter wheat (Triticum aestivum L.). The complex inheritance of resistance to SNB poses significant challenges to breeding programmes. Improving selection precision and identifying novel resistance QTLs are crucial for enhancing SNB resistance. This study investigated strategies for crop genetic improvement, including genome‐wide association mapping (GWAS) and genomic prediction (GP), within a practical breeding programme. A population of 1500 winter wheat breeding lines was phenotyped for SNB resistance over 5 years across 19 geographical locations under natural infection conditions. Despite highly unbalanced breeder's data, medium to high heritabilities for SNB resistance were achieved. GWAS identified 11 significant marker‐trait associations for SNB resistance across Chromosomes 2A, 2B, 4B, 4D, 5D and 7B. GP fivefold cross‐validation analysis revealed a predictive ability of 0.52 for SNB resistance. The resistant wheat genotypes and SNP markers identified in this study will be valuable assets for future breeding efforts to enhance SNB resistance in wheat.

Can digital financial inclusion promote the nighttime economy? - New evidence from Chinese counties’ nighttime lighting data and threshold model

In the post-pandemic era, the relative weakness of the economy is reflected not only in the macro data but also in the nighttime economy. Based on the threshold model and unbalanced panel data of 1772 county-level administrative units in China from 2014 to 2020, this paper deeply explores the relationship between digital inclusive finance and night economy and draws the following conclusions. (1) When digital financial inclusion exceeds certain thresholds, its beneficial influence on promoting the nighttime economy increases from 8.13% to 26.59% and 87.71%. (2) Digital financial inclusion promotes the nighttime economy by increasing employment in the secondary industry and promoting technological innovation. However, digital financial inclusion hinders the nighttime economy by reducing entrepreneurial activity in the accommodation and catering sectors. (3) Coverage breadth first inhibits, then promotes, and inhibits the nighttime economy in different intervals; the effect of usage depth and digital level on the nighttime economy in intervals shows an increasing trend. (4) In the Yangtze River Economic Belt region, digital financial inclusion initially inhibits the nighttime economy but turns into promotion as it develops; after the implementation of the rural revitalization strategy in 2017, digital financial inclusion can more positively and significantly promote the nighttime economy.

Beyond Sentiment Exploring the Dynamics of AIGC-Generated Sports Content and User Engagement on Xiaohongshu

This study explores the impact of AIGC-generated sports content on user sentiment and engagement on Xiaohongshu, a Chinese social media platform.Using MediaCrawler, we collected posts and comments related to sports topics generated by AI. Sentiment analysis was performed using SnowNLP to classify comments as positive or negative, with 57.1% showing positive sentiment and 42.9% negative. Despite the prevalence of positive sentiment, no significant difference in engagement (measured by likes) was observed between positive and negative comments.Temporal analysis revealed that user engagement is event-driven, with spikes in positive sentiment during major sports events. Additionally, word cloud analysis highlighted promotional and spam-like themes, suggesting challenges in content moderation. The interaction patterns showed a skewed distribution, with a small number of highly active users dominating discussions.The findings suggest that timing, relevance, and event alignment are more critical than sentiment polarity in driving engagement.The study provides insights for content creators and platforms on optimizing AIGC strategies through trend alignment and effective user collaboration.

Neurocognitive dysfunction in adolescents with recent onset major depressive disorder: a cross-sectional comparative study.

The aim of this study was to examine the neurocognitive deficits associated with the first episode of major depressive disorder (recent onset depression, ROD) in adolescents as compared to adult patients. Cross-sectional neurocognitive data from the baseline assessments of the PRONIA study with N = 650 (55.31% females) were analyzed. Based on a principal component analysis of eleven neurocognitive tests, we constructed an overall neurocognitive performance (NP) score. We examined mean score differences in NP between the groups of healthy controls (HC) and ROD and between adolescents (15-21years) and adults (22-40years) within a GLM approach. This accounts for unbalanced data with focus on interaction effects while controlling for effects of medication and educational years. Our results show lower NP for the ROD as compared to the HC group (d = -0.29, p = .046) and lower NP for the adolescent group as compared to the adult group (d = -0.29; p < .039). There was no interaction between these two group effects (F = 1.11; p = .29). Our findings suggest that the detrimental effect of ROD on neurocognitive functioning is comparable in adolescent and adult patients, since lower scores in adolescent patients are explained by effects of age and education. Neurocognitive impairment is an under addressed issue in clinical treatment guidelines for adolescent MDD. We suggest efficient monitoring in clinical practice by using an aggregate of the Digit Symbol Substitution Test and the Trail Making Test B, which highly correlated with the overall score of NP (r = 0.82).

Deep Fuzzy Credibility Surfaces for Integrating External Databases in the Estimation of Operational Value at Risk

Operational risk (OR) is usually caused by losses due to human errors, inadequate or defective internal processes, system failures, or external events that affect an organization. According to the Basel II agreement, OR is defined by seven risk events: internal fraud, external fraud, labor relations, clients, damage to fixed assets, technical failures and failures in the execution and administration of processes. However, the low frequency with which a loss event occurs creates a technological challenge for insurers in estimating the operational value at risk (OpVar) for the protection derived from an organization’s business activities. Following the above, this paper develops and analyzes a Deep Fuzzy Credibility Surface model (DFCS), which allows the integration in a single structure of different loss event databases for the estimation of an operational value at risk (OpVar), overcoming the limitations imposed by the low frequency with which a risk event occurs within an organization (sparse data). For the estimation of OpVar, the DFCS model incorporates a novel activation function based on the generalized log-logistic function to model random variables of frequency and severity that define a loss event (linguistic random variables), as well as a credibility surface to integrate the magnitude and heterogeneity of losses in a single structure as a result of the integration of databases. The stability provided by the DFCS model could be evidenced through the structure exhibited by the aggregate loss distributions (ALDs), which are obtained as a result of the convolution process between frequency and severity random variables for each database and which are expected to achieve similar structures to the probability distributions suggested by Basel II agreements (lean, long tail, positive skewness) against the OR modeling. These features make the DFCS model a reference for estimating the OpVar to protect the risk arising from an organization’s business operations by integrating internal and external loss event databases.

Analytic Moments of TGARCH(1,1) Models with Polynomially Adjusted Densities

Abstract This article extends He, Silvennoinen, and Teräsvirta (2008, J Finan Econ, 6, 208–230) and Francq and Zakoïan (2010, GARCH Models) by providing analytical expressions for the moments of the unconditional distribution of the TGARCH(1,1) under alternative specifications for the conditional mean and different skewed distributions for the innovations. We consider polynomially adjusted (PA) densities, such as the PA Logistic, PA hyperbolic secant, and the PA Gaussian, along with the skewed Student-t. Our results show that (i) the main driver of the skewness of the TGARCH(1,1) is the skewness of the innovations, while the excess kurtosis has a comparatively lesser impact. However, both skewness and kurtosis of the innovations significantly affect the TGARCH(1,1) kurtosis; (ii) if the conditional mean is not constant, returns can be asymmetric even if innovations are symmetric; (iii) skewed innovations can generate cross-correlations different from zero, indicating leverage effect, even when the volatility model is symmetric. Finally, we illustrate our theoretical results with an empirical application to stock indices.

Reliable Augmentation and Precise Identification of EPG Waveforms Based on Multi-Criteria DCGAN

The electrical penetration graph (EPG) technique is of great significance in elucidating the mechanisms of virus transmission by piercing-sucking insects and crop resistance to these insects. The traditional method of manually processing EPG signals encounters the drawbacks of inefficiency and subjectivity. This study investigated the data augmentation and automatic identification of various EPG signals, including A, B, C, PD, E1, E2, and G, which correspond to distinct behaviors exhibited by the Asian citrus psyllid. Specifically, a data augmentation method based on an improved deep convolutional generative adversarial network (DCGAN) was proposed to address the challenge of insufficient E1 waveforms. A multi-criteria evaluation framework was constructed, leveraging maximum mean discrepancy (MMD) to evaluate the similarity between the generated and real data, and singular value decomposition (SVD) was incorporated to optimize the training iterations of DCGAN and ensure data diversity. Four models, convolutional neural network (CNN), K-nearest neighbors (KNN), decision tree (DT), and support vector machine (SVM), were established based on DCGAN to classify the EPG waveforms. The results showed that the parameter-optimized DCGAN strategy significantly improved the model accuracies by 5.8%, 6.9%, 7.1%, and 7.9% for DT, SVM, KNN, and CNN, respectively. Notably, DCGAN-CNN effectively addressed the skewed distribution of EPG waveforms, achieving an optimal classification accuracy of 94.13%. The multi-criteria optimized DCGAN-CNN model proposed in this study enables reliable augmentation and precise automatic identification of EPG waveforms, holding significant practical implications for understanding psyllid behavior and controlling citrus huanglongbing.

Grandchild care and grandparents’ well-being in context: The impact of the covid-19 pandemic

Abstract Background and Objective This study investigates whether the association between supplementary grandchild care and grandparents’ subjective well-being—measured as life satisfaction, perceived stress, and loneliness—is moderated by the contextual environment. We use the Covid-19 pandemic as an example for contextual differences. Drawing on role theory, we argue that the costs and benefits of grandparenting may have differed between pandemic and pre-pandemic times. On the one hand, providing grandchild care during the pandemic may have been particularly stressful, prompting more negative effects on well-being. On the other hand, grandchild care may have been particularly relevant for enhancing well-being, as it protected grandparents from social isolation. Moreover, the association between grandparenting and well-being may have differed by gender. Research Design and Methods Using unbalanced panel data from the German Ageing Survey (DEAS) 2014 (n = 3,619), 2017 (n = 2,458), and 2020 (n = 2,021), we applied maximum likelihood structural equation modeling (ML-SEM)—a method that combines dynamic panel modeling with fixed-effects analysis—to examine whether there were differences in the relationship between grandchild care and grandparents’ well-being when comparing pandemic and pre-pandemic times and by grandparents’ gender. Results Grandchild care was associated with lower loneliness for both grandmothers and grandfathers. For grandfathers, this association was even stronger during the pandemic. Grandmothers experienced higher life satisfaction when taking care of grandchildren during the pandemic, but there was no evidence that grandchild care increased perceived stress for either grandmothers or grandfathers. Discussion and Implications In line with role enhancement theory, this study highlights that supplementary grand child care can be beneficial for grandparents’ well-being. Moreover, the context in which grandchild care takes place shapes the costs and rewards associated with it. Our results suggest that supportive policies and programs facilitating grandchild care can enhance grandparents' well-being, especially in challenging contexts.

Coinciding Diabetic Retinopathy and Diabetic Macular Edema Grading With Rat Swarm Optimization Algorithm for Enhanced Capsule Generation Adversarial Network.

In the worldwide working-age population, visual disability and blindness are common conditions caused by diabetic retinopathy (DR) and diabetic macular edema (DME). Nowadays, due to diabetes, many people are affected by eye-related issues. Among these, DR and DME are the two foremost eye diseases, the severity of which may lead to some eye-related problems and blindness. Early detection of DR and DME is essential to preventing vision loss. Therefore, an enhanced capsule generation adversarial network (ECGAN) optimized with the rat swarm optimization (RSO) approach is proposed in this article to coincide with DR and DME grading (DR-DME-ECGAN-RSO-ISBI 2018 IDRiD). The input images are obtained from the ISBI 2018 unbalanced DR grading data set. Then, the input fundus images are preprocessed using the Savitzky-Golay (SG) filter filtering technique, which reduces noise from the input image. The preprocessed image is fed to the discrete shearlet transform (DST) for feature extraction. The extracting features of DR-DME are given to the ECGAN-RSO algorithm to categorize the grading of DR and DME disorders. The proposed approach is implemented in Python and achieves better accuracy by 7.94%, 36.66%, and 4.88% compared to the existing models, such as the combined DR with DME grading for the cross-disease attention network (DR-DME-CANet-ISBI 2018 IDRiD), category attention block for unbalanced grading of DR (DR-DME-HDLCNN-MGMO-ISBI 2018 IDRiD), combined DR-DME classification with a deep learning-convolutional neural network-based modified gray-wolf optimizer with variable weights (DR-DME-ANN-ISBI 2018 IDRiD).

Infrared temperature distribution characteristics and state assessment method of sandstone under tension and compression stress

Abnormalities in infrared radiation temperature (IRT) often accompany the loading In this investigation, stress experiments and splitting tests were conducted on sandstone to investigate the IRT characteristics of sandstone under various loading conditions. The results shows that the Maximum Infrared Radiation Temperature (MAXIRT) increases by up to 5 °C, while in splitting tests, the minimum infrared radiation temperature (MINIRT) decreases by approximately 1.5 °C. Compared to the average infrared radiation temperature (AIRT), MAXIRT is more sensitive to compressive stress, whereas (MINIRT) is more responsive to tensile stress. The correlation between IRT indicators and stress were further analyzed. The findings reveal a positive correlation between IRT and compressive stress. At low stress levels, the correlation coefficient fluctuates between −0.4 and 0.6, generally showing low correlation. As stress increases, the correlation coefficient rises, reaching above 0.8, indicating a high correlation. Tensile stress exhibits a negative correlation, with a consistent trend. Additionally, a statistical analysis of the frequency distribution of IRT under different stress conditions was conducted, followed by hypothesis testing. The results demonstrate that the probability distribution of IRT during loading follows a skewed normal distribution. The changes in skewness can be divided into three stages: initial fluctuation, stable variation, and abrupt failure. Under compressive stress, a right-skewed distribution is observed, with skewness exceeding 3.5 before fracture and reaching above 10 near fracture. Under tensile stress, a left-skewed distribution is noted, with skewness reaching −4.5 near fracture. These findings contribute to the assessment of rock stress states and provide early warning information for predicting rock failure.

Seasonal Characteristics of Particulate Matter by Pollution Source Type and Urban Forest Type

To provide consistent air purification benefits from urban forests, it is crucial to identify common characteristics that allow for similar experimental setups. This study aimed to analyze PM10 concentrations in urban forests near pollution sources and understand their mitigation effects. Data from the Asian Initiative for Clean Air Networks, Korea, were used, focusing on three urban forests adjacent to road and industrial pollution sources in Korea, with PM10 concentrations collected during 2021. Considering high PM10 concentrations during winter and spring, these seasons were divided into two sub-periods, resulting in six seasonal periods for analysis. To address the right-skewed PM10 distribution and reduce outlier influence, the Kruskal–Wallis test was used. The results showed that “good” PM10 levels were lowest in early spring, increasing to a peak in summer before declining. High PM10 events were concentrated in spring, early spring, and early winter. The Kruskal–Wallis test indicated lower median PM10 concentrations in urban forests compared to pollution sources in the latter half of the year, while no significant median differences were found in the first half. Distribution visualizations further confirmed that even during high PM10 periods, all urban forests showed lower PM10 values compared to pollution sources. In conclusion, PM10 concentrations in urban forests were consistently lower than in pollution sources across all seasons, demonstrating their effectiveness in air purification at both road and industrial pollution sources. Future research should consider additional variables, such as PM2.5, to further explore differences between pollution sources.

Enhancing Bucket Sort Efficiency: A Combinatorial Approach to Algorithm Optimization

Abstract: The efficiency of sorting algorithms is critical to numerous computational processes, including data management, search operations, and real-time applications. Among these algorithms, Bucket Sort is notable for its linear-time performance when data is uniformly distributed over a defined range. However, its efficiency suffers significantly with non-uniform data distributions, resulting in unbalanced buckets and increased sorting time. In our research, we explore how combinatorial algorithms can enhance Bucket Sort’s efficiency by optimizing bucket partitioning, dynamic allocation, and internal sorting strategies. Through rigorous mathematical proofs and analysis, we demonstrate that the application of combinatorial techniques yields substantial improvements in average-case performance, particularly in handling skewed data distributions. Practical case studies in large-scale data processing illustrate the adaptability and effectiveness of these optimizations, supporting their potential in real-world applications.

Comparison and analysis of CNN models to Address Skewed Data Issues in Alzheimer's Diagnosis

Comparison and analysis of CNN models to Address Skewed Data Issues in Alzheimer's Diagnosis

Prediction Intervals for Overdispersed Poisson Data and Their Application in Medical and Pre-Clinical Quality Control.

In pre-clinical and medical quality control, it is of interest to assess the stability of the process under monitoring or to validate a current observation using historical control data. Classically, this is done by the application of historical control limits (HCL) graphically displayed in control charts. In many applications, HCL are applied to count data, for example, the number of revertant colonies (Ames assay) or the number of relapses per multiple sclerosis patient. Count data may be overdispersed, can be heavily right-skewed and clusters may differ in cluster size or other baseline quantities (e.g., number of petri dishes per control group or different length of monitoring times per patient). Based on the quasi-Poisson assumption or the negative-binomial distribution, we propose prediction intervals for overdispersed count data to be used as HCL. Variable baseline quantities are accounted for by offsets. Furthermore, we provide a bootstrap calibration algorithm that accounts for the skewed distribution and achieves equal tail probabilities. Comprehensive Monte-Carlo simulations assessing the coverage probabilities of eight different methods for HCL calculation reveal, that the bootstrap calibrated prediction intervals control the type-1-error best. Heuristics traditionally used in control charts (e.g., the limits in Shewhart c- or u-charts or the mean ± 2 SD) fail to control a pre-specified coverage probability. The application of HCL is demonstrated based on data from the Ames assay and for numbers of relapses of multiple sclerosis patients. The proposed prediction intervals and the algorithm for bootstrap calibration are publicly available via the R package predint.

Emergency Medicine Clerkship Grading Scheme, Grade, and Rank-List Distribution as Reported on Standardized Letters of Evaluation

Background: The Standardized Letter of Evaluation (SLOE) is a crucial component of the emergency medicine (EM) application process. Given the critical role of the SLOE, we attempted to better understand the grading scales used, as well as the distribution of grades and rank-list positions.Objectives: Our primary objective in this study was to determine the distribution of grading formats, grades given, and rank-list positions across EM clerkships using the SLOE. Methods: We performed a cross-sectional study of the grading formats, grades given, and ranking distributions as reported on the SLOE during the 2022–23 application cycle. We obtained data on SLOEs from EM residency programs accredited by the Accreditation Council for Graduate Medical Education by reviewing all applicants who applied to either of two EM residency programs in geographically different regions. Trained abstractors recorded the following data: number of students rotating in the prior year; grading format used; and grade and rank distribution among students. Results: We included 264 programs in our final analysis, after 13 programs met exclusion criteria. The majority of programs (72.2%) use an Honors/High Pass/Pass/Fail grading scheme. We determined the mean percent of each grade: Honors/A 27.6%; High Pass/B 31.1%; Pass/C 40.8%; Low Pass/D 0.2%; and Fail/F 0.3%. Finally, we determined the mean percent for each rank-list position: top 10% was 17.6%; top third 36.5%; mid third 34.1%; and low third 11.8%. Conclusion: We determined the grading schemes and grade and rank-list distributions for EM programs during the 2022–2023 academic year. Most programs used a Honors/High Pass/Pass/Fail grading scheme, with the majority of students receiving Honors or High Pass, while 0.3% failed their rotation. Both grades and rank list demonstrated evidence of a skewed distribution toward higher grades and rank-list position.

Population-aware permutation-based significance thresholds for genome-wide association studies

Abstract Motivation Permutation-based significance thresholds have been shown to be a robust alternative to classical Bonferroni significance thresholds in genome-wide association studies for skewed phenotype distributions. The recently published method permGWAS introduced a batch-wise approach to efficiently compute permutation-based genome-wide association studies. However, running multiple univariate tests in parallel leads to many repetitive computations and increased computational resources. More importantly, traditional permutation methods that permute only the phenotype break the underlying population structure. Results We propose permGWAS2, an improved method that does not break the population structure during permutations and uses an elegant block matrix decomposition to optimize computations, thereby reducing redundancies. We show on synthetic data that this improved approach yields a lower false discovery rate for skewed phenotype distributions compared to the previous version and the commonly used Bonferroni correction. In addition, we re-analyze a dataset covering phenotypic variation in 86 traits in a population of 615 wild sunflowers (Helianthus annuus L.). This led to the identification of dozens of novel associations with putatively adaptive traits, and removed several likely false-positive associations with limited biological support. Availability permGWAS2 is open-source and publicly available on GitHub for download: https://github.com/grimmlab/permGWAS. Supplementary information Supplementary data are available at Bioinformatics Advances online.

Peripheral blood immune cells distribution in biopsy-proven myocarditis: etiology and prognostic correlates

Abstract Background Myocarditis is an inflammatory disease of the myocardium with viral or immune-mediated/autoimmune etiology. Definite etiological diagnosis relays on endomyocardial biopsy (EMB). High titre anti-heart autoantibodies (AHA) define severe autoimmune forms. Virus-negative myocarditis may require immunosuppression (IS) to reduce progression to dilated cardiomyopathy, heart transplant or death. Prognostic stratification is incomplete and IS is not always efficacious. Animal and human studies suggest a pathogenetic role of immune cells, but little is known on their peripheral distribution or prognostic role. Purpose Characterize EMB-proven myocarditis patients’ peripheral blood to identify new non-invasive etiological and prognostic biomarkers. Methods Seventy-eight EMB-proven myocarditis patients and 9 healthy controls (HC) were enrolled according to the Declaration of Helsinki and written informed consent was collected for each participant. Peripheral blood mononuclear cells were purified by Ficoll stratification and immune cells distribution was evaluated by flow cytometry. Variables were analysed by Mann-Whitney or Kruskall-Wallis and Dunn post-hoc tests clustering patients according to clinical features, EMB and AHA results. Results Compared to HC (figure 1), plasmacytoid dendritic cells were reduced in myocarditis patients: with autoimmune/lymphocytic forms (p=0.0091); without extra cardiac autoimmune diseases (AD, p=0.0075); in those unresponsive to IS (p=0.0379) or never treated because of spontaneous healing (p=0.0015). Moreover, several immunophenotypical features discriminated myocarditis type/IS-response from HC. Viral forms were characterized by reduced CD94+ NK cells and CCR2 over-expression in intermediate monocytes (p=0.029, p=0.013). Autoimmune cases were defined by: higher CD62L+ NK cells when AHA negative (p=0.019); increased Th1 if not requiring IS (p=0.03); higher γδ-TCR+ Th17 and reduced Treg cells if without other AD (both p=0.04), or higher CD4+/IL17+ cells with concurrent AD (p=0.02) and AHA positivity (p=0.036). Moreover, ongoing IS treatment influenced peripheral cells distribution: whole blood cell counts were more frequently altered and total NK cells were reduced (p=0.007); treg cells were reduced in patients actively treated but unresponsive to IS (p=0.036). Conclusion Biopsy-proven myocarditis patients showed a skewed peripheral blood distribution of either innate or adaptive immune cells according to different histology, etiology (viral vs autoimmune) and response to IS in autoimmune forms, unveiling potential new non-invasive immunological biomarkers for myocarditis. Figure: Plasmacytoid dendritic cells (pDCs) percentage was assessed by flow-cytometry and data were showed by boxplots clustering myocarditis patients for etiology (A), EMB histological result (B), presence of extracardiac AD (C) or IS response (D). Data were analysed by Kruskall-Wallis (graph bottom) and Dunn post-hoc tests.

HADNet: A Novel Lightweight Approach for Abnormal Sound Detection on Highway Based on 1D Convolutional Neural Network and Multi-Head Self-Attention Mechanism

Video surveillance is an effective tool for traffic management and safety, but it may face challenges in extreme weather, low visibility, areas outside the monitoring field of view, or during nighttime conditions. Therefore, abnormal sound detection is used in traffic management and safety as an auxiliary tool to complement video surveillance. In this paper, a novel lightweight method for abnormal sound detection based on 1D CNN and Multi-Head Self-Attention Mechanism on the embedded system is proposed, which is named HADNet. First, 1D CNN is employed for local feature extraction, which minimizes information loss from the audio signal during time-frequency conversion and reduces computational complexity. Second, the proposed block based on Multi-Head Self-Attention Mechanism not only effectively mitigates the issue of disappearing gradients, but also enhances detection accuracy. Finally, the joint loss function is employed to detect abnormal audio. This choice helps address issues related to unbalanced training data and class overlap, thereby improving model performance on imbalanced datasets. The proposed HADNet method was evaluated on the MIVIA Road Events and UrbanSound8K datasets. The results demonstrate that the proposed method for abnormal audio detection on embedded systems achieves high accuracy of 99.6% and an efficient detection time of 0.06 s. This approach proves to be robust and suitable for practical applications in traffic management and safety. By addressing the challenges posed by traditional video surveillance methods, HADNet offers a valuable and complementary solution for enhancing safety measures in diverse traffic conditions.

A Study on Predicting Traffic Flow by Random Forest Based on Treatment of Eliminating Skewed Distribution

A Study on Predicting Traffic Flow by Random Forest Based on Treatment of Eliminating Skewed Distribution