Recall Index Research Articles

Land Use and Land Cover Change Analysis in Brazilian Urban centres: an approach using Cellular Automata and Neural Networks

Abstract. During the 1970s, Brazil experienced most of its population living in cities for the first time. Historically, urbanization has been intensely related to land use and cover changes, amplifying climatic and ecological stress in recently established metropolises. This study aims to analyze, through classified Remote Sensing images and the Cellular Automata and Artificial Neural Networks (CA-ANN) machine learning model, land use and land cover trends in regions that have experienced accentuated demographic growth in the decades 2000 and 2010. The methodology consists of: i) identifying areas with a high density of buildings, ii) defining the variables that drive land use change, and iii) proposing a methodology for predicting changes in the urban area. The results indicate that the urban class prediction presented high precision (≥ 0.74) and recall (≥ 0.86) indices. Forest class also presented a high precision score (≥0.72), showing an elevated prediction hit rate. Furthermore, the proposed methodology improved the results obtained in previous works for the same cities, presenting higher Kappa values in all cases.

The relationship between childhood SES and health in middle and old age: evidence from China.

Drawing from the extended Grossman health capital and demand theory and the life course theory, this study examined whether childhood SES has direct and significant correlation with health in middle and old age in a specific historical context in China. A sample of 9,861 respondents was obtained from the China Health and Retirement Longitudinal Study (CHARLS). Childhood SES was measured by objective indices of recall. Health was assessed by self-reported, physician diagnosis and the Center for Epidemiologic Studies Depression Scale (CESD). The Propensity Score Matching (PSM) was used to estimate the treatment effect between childhood SES and later life health. The Karlson-Holm-Breen (KHB) method was employed to examine the associative mediation effects. Compared to respondents with low SES in childhood, respondents with high SES in childhood had, on average, 5.1% more likely to report their health as good, an average 2.4% lower prevalence of chronic diseases and an average 7.6% lower in the score of depression in middle and old age. The indirect relationships of childhood health, adulthood SES and adulthood lifestyle with health in middle and old age were all significant. SES upward mobility in adulthood can diminish the association between childhood disadvantage and poor health in middle and old age. The health effects of childhood SES can persist into middle and old age, this is more noticeable in rural areas, particularly in females. The critical period, cumulative risk and social mobility models produce synergistic effects in China. Our results also promote a paradigm shift in health interventions from old age to early life for health-vulnerable populations.

Detection Method of Stator Coating Quality of Flat Wire Motor Based on Improved YOLOv8s.

The stator of a flat wire motor is the core component of new energy vehicles. However, detecting quality defects in the coating process in real-time is a challenge. Moreover, the number of defects is large, and the pixels of a single defect are very few, which make it difficult to distinguish the defect features and make accurate detection more difficult. To solve this problem, this article proposes the YOLOv8s-DFJA network. The network is based on YOLOv8s, which uses DSFI-HEAD to replace the original detection head, realizing task alignment. It enhances joint features between the classification task and localization task and improves the ability of network detection. The LEFG module replaces the C2f module in the backbone of the YOLOv8s network that reduces the redundant parameters brought by the traditional BottleNeck structure. It also enhances the feature extraction and gradient flow ability to achieve the lightweight of the network. For this research, we produced our own dataset of stator coating quality regarding flat wire motors. Data augmentation technology (Gaussian noise, adjusting brightness, etc.) enriches the dataset, to a certain extent, which improves the robustness and generalization ability of YOLOv8s-DFJA. The experimental results show that in the performance of YOLOv8s-DFJA compared with YOLOv8s, the mAP@.5 index increased by 6.4%, the precision index increased by 1.1%, the recall index increased by 8.1%, the FPS index increased by 9.8FPS/s, and the parameters decreased by 3 Mb. Therefore, YOLOv8s-DFJA can be better realize the fast and accurate detection of the stator coating quality of flat wire motors.

Improved electrical coupling integrated energy system based on particle swarm optimization

The rational utilization of energy is an important issue for sustainable development. Electrically coupled integrated energy systems can enhance energy utilization efficiency and reduce energy costs. However, traditional integrated energy system optimization has problems with local optima and slow convergence speed, which cannot fully utilize energy resources. Therefore, this study proposes an improved electrical coupling integrated energy system on the ground of particle swarm optimization algorithm. In response to the problems of local optima and slow convergence speed in traditional optimization algorithms, particle swarm optimization algorithm is introduced for system optimization. By combining PSO with simulated annealing algorithm, the possibility of PSO in global optimization is reduced. The local search ability of PSO and the global search ability of simulated annealing algorithm are used to find the optimal solution. The particle swarm optimization algorithm is used for preliminary search. When the particle falls into the local optimal, the simulated annealing algorithm is introduced for global search, and the particle is guided to jump out of the local optimal and continue searching. The experiment demonstrates that the improved algorithm has certain advantages in solving test functions. The variance, mean, and optimal values are 0.00125, 0.13874, and 0.105531, respectively, which are all better than the particle swarm optimization algorithm. The simulated annealing algorithm improved the particle swarm optimization algorithm with a high accuracy index, which eventually stabilized above 0.9, and the recall index also remained above 0.8. After 100 iterations, it had already fallen into a local optimal solution. By applying the improved hybrid optimization algorithm to the electrically-coupled integrated energy system, the distribution of various energy sources can be managed and optimized more effectively, and the overall performance of the system can be improved. Especially when dealing with complex energy scheduling and distribution problems, the algorithm can provide more stable, efficient and reliable solutions. This study can achieve efficient operation and optimized scheduling of integrated energy systems, reduce energy consumption and environmental pollution, and reduce energy costs. And it can improve the reliability and stability of energy supply, which has important application value and research significance.

Deep-Learning- and Unmanned Aerial Vehicle-Based Structural Crack Detection in Concrete

Deep-learning- and unmanned aerial vehicle (UAV)-based methods facilitate structural crack detection for tall structures. However, contemporary datasets are generally established using images taken with handheld or vehicle-mounted cameras. Thus, these images might be different from those taken by UAVs in terms of resolution and lighting conditions. Considering the difficulty and complexity of establishing a crack image dataset, making full use of the current datasets can help reduce the shortage of UAV-based crack image datasets. Therefore, the performance evaluation of existing crack image datasets in training deep neural networks (DNNs) for crack detection in UAV images is essential. In this study, four DNNs were trained with different architectures based on a publicly available dataset and tested using a small UAV-based crack image dataset with 648 +pixel-wise annotated images. These DNNs were first tested using the four indices of precision, recall, mIoU, and F1, and image tests were also conducted for intuitive comparison. Moreover, a field experiment was carried out to verify the performance of the trained DNNs in detecting cracks from raw UAV structural images. The results indicate that the existing dataset can be useful to train DNNs for crack detection from UAV images; the TransUNet achieved the best performance in detecting all kinds of structural cracks.

Plasma biomarkers, memory impairment, and imaging measures of Alzheimer’s disease in American Indians

AbstractBackgroundAlzheimer’s disease (AD) is best defined in vivo using quantifiable, objective markers reflecting amyloid deposition (A), pathologic tau (T), and neurodegeneration (N). However, gold standards for ATN (imaging or cerebrospinal proteins) are expensive, invasive, and/or impractical in low resource settings. Circulating proteins offer a practical alternative. Candidate plasma markers include amyloid‐β (Aβ40&42), phosphorylated tau‐181 (ptau), glial fibrillary protein (GFAP), and neurofilament light chain (NfL). However, no such markers have been evaluated in American Indians.MethodThe Strong Heart Study recruited 401 American Indians over 3 states in 2017‐2019 for MRI, cognitive testing, and clinical examination. MRI included T1 sequences, which were processed using FreeSurfer to quantify structural volumes. Neuropsychological testing incuded California Verbal Learning Test, which was operationalized into categories of memory impairment using indices of learning and recall. Plasma samples were sent to University of Gothenburg for Simoa Quanterix assay. Statistics included descriptive summary; multivariate regression; and receiver operating characteristic analysis.ResultThe study population was mean age 78 (SD 4.7, range 70‐95); majority female (71%); and 21% APOE ε4 allele carriers. An estimated 31% had any memory impairment (13.8% retrieval, 6.5% storage, 10.7% encoding). Median plasma marker values were: ptau 5.0pg/mL (range: 1.5‐442), Aβ40 128pg/mL (116‐175), Aβ42 8.2pg/mL (6.7‐9.7), GFAP 150pg/mL (0‐651), and NfL 31.6pg/mL (9.7‐343); median Aβ42:40 ratio 0.06. Comparing APOEε4 carriers to non‐carriers (Table1), ptau was +12%, Aβ40 ‐0.8%, Aβ42 ‐8.4%, Aβ4240 ‐16.7% (P = 0.006), GFAP +16.1%, NfL +3.2%, but only Aβ4240 was significant. Linear regressions with brain volumes identified significant associations for Aβ42, Aβ40, GFAP, and NfL (but not ptau) with ventricle, cortical, and overall volume; NfL was also associated with entorhinal and hippocampal volume. GFAP and ptau were significantly higher in those with memory encoding deficit; NfL higher with encoding or retrieval deficit (Table2). Although not significant, Aβ40 was lower with storage deficit, and Aβ42 with encoding or storage deficit. AUC range for plasma markers to discriminate memory impaired from intact: 0.4‐0.6.ConclusionThese data are a first look at ATN plasma markers in a population‐based cohort of American Indians, an underrepresented group in AD research. Future research should address generalizability and measurement validity.

75 The Association Between Cognitive Function and Older Adults Performance on a Naturalistic Cooking Task in the Home Environment

Objective:Cognitive impairment can affect an individual’s ability to perform routine tasks. In this study, we investigate how cognitive abilities relate to the accuracy and efficiency of performance on a naturalistic cooking task completed in older adults’ home environments. We hypothesized a positive association between task accuracy and global cognitive status, and task efficiency and executive functioning. We further hypothesized a negative association between omission errors and immediate and delayed memory recall.Participants and Methods:Fourteen community-dwelling older adults (Age, M = 73.92 years; Female = 9; Education, M = 16.38 years) along the continuum from normal aging to mild dementia completed a “Cooking Task” in their home environment. Specifically, participants were instructed to fry or scramble an egg, prepare slice of toast with jelly, serve side of sliced apple, pour glass of water, bring prepared items to table, and clean dishes used. Participants received ingredients necessary for task completion and a task list to reference. The task efficiency score (range 0-6) was based on multi-tasking and organizational skills (e.g., beginning the egg task early in session, plating items as prepared). Overall accuracy was computed by identifying error types (e.g., inefficiencies, substitutions, omissions, and subtasks attempted) and scaling accuracy (range 1-5) for each subtask, then summing all six subtask accuracy scores to get overall accuracy (range 6-30). Participants also completed a range of neuropsychological assessments, which included the Telephone Interview for Cognitive Status, Letter and Category Fluency from the Delis-Kaplan Executive Function System, and immediate and delayed recall measures from the Repeatable Battery for the Assessment of Neuropsychological Status. Due to the small sample size, findings are preliminary, and scatterplots were evaluated for outliers that might influence findings.Results:Consistent with hypotheses, as overall accuracy on the Cooking Task increased so did performance on the global cognitive measure (TICS: r = 0.61, p = 0.02). Lower rates of omission errors were also associated with better performance on both immediate (r = -0.75, p < 0.01) and delayed (r = -0.55, p = 0.04) recall indices. However, these findings were not specific, as overall accuracy also significantly correlated with the memory indices and verbal fluency measures (ps < 0.05). Additionally, lower rates of omission errors significantly correlated with performance on the TICS and the D-KEFS Letter Fluency (ps < 0.05). Contrary to our hypothesis, no significant associations were found between cooking task efficiency and executive functioning (D-KEFS subtests). There were also no significant correlations between cooking task efficiency and global cognitive status or memory.Conclusions:The present study supported our hypotheses that better overall task accuracy is associated with higher cognitive status and lower rates of omission errors correlate with better immediate and delayed recall abilities. However, the findings were not specific to these domains of functioning but rather suggest that clinical assessments measuring a range of cognitive abilities are related to the accuracy of daily task performance and omission errors on routine daily tasks. Future research will explore the validity of the efficiency measure.

Detection of fetal facial anatomy in standard ultrasonographic sections based on real-time target detection network.

At present, prenatal ultrasound is one of the important means for screening fetal malformations. In the process of prenatal ultrasound diagnosis, the accurate recognition of fetal facial ultrasound standard plane is crucial for facial malformation detection and disease screening. Due to the dense distribution of fetal facial images, no obvious structure contour boundary, small structure area, and large area overlap in the middle of the structure detection frame, this paper regards the fetal facial standard plane and its structure recognition as a universal target detection task for the first time, and applies real-time YOLO v5s to the fetal facial ultrasound standard plane structure detection and classification task. First, we detect the structure of a single slice, and take the structure of a slice class as the recognition object. Second, we carry out structural detection experiments on three standard planes; then, on the basis of the previous stage, the images of all parts included in the ultrasound examination of multiple fetuses were collected. In the single-class structure detection experiment and the structure detection and classification experiment of three types of standard planes, the overall recognition effect of Precision and Recall index data is better, with Precision being 98.3% and 98.1%, and Recall being 99.3% and 98.2%, respectively. The experimental results show that the model has the ability to identify fetal facial anatomy and standard sections in different data, which can help the physician to automatically and quickly screen out the standard sections of each fetal facial ultrasound.

A POI Recommendation Model for Intelligent Systems Using AT-LSTM in Location-Based Social Network Big Data

In location-based social networks (LBSN), users can check-in at points of interest (POI) to record their trips. POI recommendation is an important service provided by LBSN; it can help users quickly find POI of interest, and also help POI providers more comprehensively understand user preferences and improve service quality. This paper proposes a POI recommendation algorithm that is based on attention mechanism. The sequence characteristics and short-term preferences of historical data are captured through the attention mechanism module and long short-term memory network (LSTM), and the POI location prediction is carried out in combination with the user embedding characteristics, and a better prediction accuracy is obtained. These results simulated show that the proposed method can realize the reliable analysis of complex data sets, and its precision index remains above 0.1 and recall index remains above 0.08, and it can also alleviate the cold start problem and better meet the personalized needs of users.

Cognitive aging in rural communities: preliminary memory characterization of a community cohort from Southern Nevada.

Rural-dwelling older adults face unique health challenges that may increase risk for Alzheimer's disease and dementia but are underrepresented in aging research. Here, we present an initial characterization of a rural community cohort compared to an urban cohort from the same region. Adults over age 50 living in a non-metropolitan area are clinically characterized using the Uniform Data Set, enriched with additional measures of verbal and non-verbal memory measures. Neighborhood disadvantage is also assessed. Clinical and cognitive differences between cohorts were explored after stratifying by cognitive impairment. Between group comparisons found that rural-dwellers demonstrated better verbal memory than urban-dwellers on primary indices of learning, recall, and recognition, with small to medium effects in overall comparisons. When stratified by impairment, rural-urban differences were notably larger among cognitively normal individuals. Within-group comparisons found that the magnitude of impairment between cognitively normal and impaired groups was greater among rural-dwellers compared to urban-dwellers. No differences in non-verbal memory or overall clinical status were found, and there were no effects of neighborhood disadvantage on any cognitive measure. Living in a rural community presents a complex set of contextual factors that for some, may increase risk for dementia. In this study, we found small to moderate memory advantages for rural-dwellers, leaving open the possibility that late-life rural living may be advantageous for some and promote resilience. Additional prospective research is critically needed to better understand the factors that influence aging outcomes in this underrepresented population.

Research on the blended teaching path of water conservancy engineering in universities based on big data analysis

Abstract The rapid development of big data technology brings opportunities and challenges to traditional teaching methods, and online teaching mode is gradually carried out widely and deeply in colleges and universities. This paper proposes a framework for constructing students’ online learning evaluation process by using big data methods. The focus is on data analysis of the evaluation model of water resources engineering in colleges and universities through the BP algorithm. Under the accuracy index, the BP algorithm 93% performs better than the other three algorithms. The higher the accuracy rate, the more it reflects the feasibility and correctness of the model. In the recall index, the recall rate of the BP algorithm fluctuates from 80% to 93%, the fluctuation range of 13% is more stable, and the overall performance is better than the other three algorithms. The objective evaluation of learning obtained through big data will greatly reduce teachers’ teaching burden and produce more hydraulic engineering professionals for China.

DCF-Yolov8: An Improved Algorithm for Aggregating Low-Level Features to Detect Agricultural Pests and Diseases

The invasion of agricultural diseases and insect pests is a huge difficulty for the growth of crops. The detection of diseases and pests is a very challenging task. The diversity of diseases and pests in terms of shapes, colors, and sizes, as well as changes in the lighting environment, have a massive impact on the accuracy of the detection results. We improved the C2F module based on DenseBlock and proposed DCF to extract low-level features such as the edge texture of pests and diseases. Through the sensitivity of low-level features to the diversity of pests and diseases, the DCF module can better cope with complex detection tasks and improve the accuracy and robustness of the detection. The complex background environment of pests and diseases and different lighting conditions make the IP102 data set have strong nonlinear characteristics. The Mish activation function is selected to replace the CBS module with the CBM, which can better learn the nonlinear characteristics of the data and effectively solve the problems of gradient disappearance in the algorithm training process. Experiments show that the advanced Yolov8 algorithm has improved. Comparing with Yolov8, our algorithm improves the MAP50 index, Precision index, and Recall index by 2%, 1.3%, and 3.7%. The model in this paper has higher accuracy and versatility.

Local performance evaluation of AI-algorithms with the generalized spatial recall index

Abstract We have developed a novel metric to gauge the performance of artificial intelligence (AI) or machine learning (ML) algorithms, called the Spatial Recall Index (SRI). The novelty is the spatial resolution of a standard performance indicator, as a Recall value is assigned to each individual pixel. This generates a distribution of the performance of a given AI-algorithm with the resolution of the images in the dataset. While the mathematical basis has already been presented before, here we demonstrate the usage on more datasets and delve into in-depth application examples. We examine both the MS COCO and the Berkeley Deep Drive datasets, using a state-of-the-art object detection algorithm. The dataset is degraded using a physical-realistic lens-model, where the optical performance varies over the field of view, as a real camera would. This study highlights the usefulness of the SRI, as every image has been taken by realistic optics. A generalization, the GSRI is introduced, from which we derive SRIA, weighting with object area and SRIrisk intended for autonomous driving. Finally, these metrics are compared.

Estimation of non-alcoholic steatohepatitis (NASH) disease using clinical information based on the optimal combination of intelligent algorithms for feature selection and classification

The early diagnosis of NASH disease can decrease the risk of proceeding elements and treatment costs for patients. This study aims to present an optimal combination of intelligent algorithms using advanced machine learning methods, including different feature selections and classifications based on clinical data and blood factors. In this work, collected data were from 176 patients to investigate NASH disease, and 19 features were extracted. We then sought to find the best combination of features based on different feature selection algorithms such as Feature Forward Selection (FFS), Minimum Redundancy Maximum Relevance (MRMR), and Mutual Information (MI). Finally, we used nine classifier frameworks with different mathematical mechanisms, including random forest (RF), logistic regression (LR), Linear Discriminant Analysis (LDA), AdaBoost, K nearest neighbors (KNN), multilayer perceptron model (MLP), support vector machine (SVM), and decision tree (DT) to estimate NASH disease. Our investigation revealed that the combination of dominant features, namely body mass index (BMI), glutamic pyruvic transaminase (GPT), total cholesterol (TC), high-density lipoprotein (HDL), Ezetimibe, lipoprotein level Lp(a), Loge(Lp(a)), total triglyceride (TG), Creatinine (Cre), HbA1c, Fibrate, and Sex, selected by the MRMR algorithm and classified by the RF method can provide the most appropriate performance based on less computation effort and maximum performance with accuracy, AUC, precision, and recall indices, which are and respectively. This study investigated the configuration of feature selection and classifier that is most suitable for classifying NASH disease based on clinical data and blood factors. The proposed intelligent algorithm based on MRMR and RF classifier can automatically diagnose NASH disease with appropriate performance and present an initial report without any further invasive methods. It also clarifies the diagnostic process and, as a result, the continuation of their prevention and treatment cycle.

Recall of substandard medicines in Brazil during the period 2010–2018

BackgroundEven with all the care taken during the production process, the pharmaceutical industries are still subject to manufacturing medicines with quality deviations, generating commercialized products without the required quality and necessitating their subsequent recall from the market. The objective of this study was to evaluate the reasons that led to the recall of medicines in Brazil in the period evaluated.MethodsThis is a descriptive study (using document analysis), on the recall of substandard medicines registered on the website of the National Health Surveillance Agency (ANVISA), from 2010 to 2018. The variables studied were the type of medicine (reference, generic, similar, specific, biological, herbal, simplified notification, new and radiopharmaceutical), type of pharmaceutical dosage form (solid, liquid, semi-solid and parenteral preparation), and reason for recall (Good manufacturing practices, quality and quality/good manufacturing practices).ResultsA total of n = 3,056 recalls of substandard medicine were recorded. Similar medicines had a higher recall index (30.1%), followed by generics (21.3%), simplified notification (20.7%) and reference (12.2%). Different dosage forms had similar recalls: solids (35.2%), liquids (31.2%) and parenteral preparations (30.0%), with the exception of semi-solids (3.4%). The reasons for the highest occurrences were related to good manufacturing practices (58.4%) and quality (40.4%).ConclusionThe probable cause for this high number of recalls is the fact that, even with all the quality controls and processes in accordance with good manufacturing practices, errors can occur, both human and in automated processes, thus causing the release of batches that should not have been approved. In summary, it is necessary for manufacturers to implement a robust and well structured quality system in order to avoid such deviations, and it is up to ANVISA to apply greater oversight in the post marketing of these products.

Forecasting of Post-Graduate Students’ Late Dropout Based on the Optimal Probability Threshold Adjustment Technique for Imbalanced Data

The purpose of this research article was to contrast the benefits of the optimal probability threshold adjustment technique with other imbalanced data processing techniques, in its application to the prediction of post-graduate students’ late dropout from distance learning courses in two universities in the Ibero-American space. In this context, the optimization of the Logistic Regression, Random Forest, and Neural Network classifiers, together with different techniques, attributes, and algorithms (Hyperparameters, SMOTE, SMOTE_SVM, and ADASYN) resulted in a set of metrics for decision-making, prioritizing the reduction of false negatives. The best model was the Neural Network model in combination with SMOTE_SVM, obtaining a recall index of 0.75 and an f1-Score of 0.60. Likewise, the robustness of the Random Forest classifier for imbalanced data was demonstrated by achieving, with an optimal threshold of 0.427, very similar metrics to those obtained by the consensus of the three best models found. This demonstrates that, for Random Forest, the optimal prediction probability threshold is an excellent alternative to resampling techniques with different optimal thresholds. Finally, it is hoped that this research paper will contribute to boost the application of this simple but powerful technique, which is highly underrated with respect to data resampling techniques for imbalanced data.

Spatio-Temporal Transformer Recommender: Next Location Recommendation with Attention Mechanism by Mining the Spatio-Temporal Relationship between Visited Locations

Location-based social networks (LBSN) allow users to socialize with friends by sharing their daily life experiences online. In particular, a large amount of check-ins data generated by LBSNs capture the visit locations of users and open a new line of research of spatio-temporal big data, i.e., the next point-of-interest (POI) recommendation. At present, while some advanced methods have been proposed for POI recommendation, existing work only leverages the temporal information of two consecutive LBSN check-ins. Specifically, these methods only focus on adjacent visit sequences but ignore non-contiguous visits, while these visits can be important in understanding the spatio-temporal correlation within the trajectory. In order to fully mine this non-contiguous visit information, we propose a multi-layer Spatio-Temporal deep learning attention model for POI recommendation, Spatio-Temporal Transformer Recommender (STTF-Recommender). To incorporate the spatio-temporal patterns, we encode the information in the user’s trajectory as latent representations into their embeddings before feeding them. To mine the spatio-temporal relationship between any two visited locations, we utilize the Transformer aggregation layer. To match the most plausible candidates from all locations, we develop on an attention matcher based on the attention mechanism. The STTF-Recommender was evaluated with two real-world datasets, and the findings showed that STTF improves at least 13.75% in the mean value of the Recall index at different scales compared with the state-of-the-art models.

A Novel Deep Learning Method for Automatic Recognition of Coseismic Landslides

Massive earthquakes generally trigger thousands of coseismic landslides. The automatic recognition of these numerous landslides has provided crucial support for post-earthquake emergency rescue, landslide risk mitigation, and city reconstruction. The automatic recognition of coseismic landslides has always been a difficult problem due to the relatively small size of a landslide and various complicated environmental backgrounds. This work proposes a novel semantic segmentation network, EGCN, to improve the landslide identification accuracy. EGCN conducts coseismic landslide recognition by a recognition index set as the input data, CGBlock as the basic module, and U-Net as the baseline. The CGBlock module can extract the relatively stable global context-dependent features (global context features) and the unstable local features by the GNN Branch and CNN Branch (GNN Branch contains the proposed EISGNN) and integrates them via adaptive weights. This method has four advantages. (1) The recognition indices are established according to the causal mechanism of coseismic landslides. The rationality of the indices guarantees the accuracy of landslide recognition. (2) The module of EISGNN is suggested based on the entropy importance coefficient and GATv2. Owing to the feature aggregation among nodes with high entropy importance, global and useful context dependency can be synthesized and the false alarm of landslide recognition can be reduced. (3) CGBlock automatically integrates context features and local spatial features, and has strong adaptability for the recognition of coseismic landslides located in different environments. (4) Owing to CGBlock being the basic module and U-Net being the baseline, EGCN can integrate the context features and local spatial characteristics at both high and low levels. Thus, the accuracy of landslide recognition can be improved. The meizoseismal region of the Ms 7.0 Jiuzhaigou earthquake is selected as an example to conduct coseismic landslide recognition. The values of the precision indices of Overall Accuracy, mIoU, Kappa, F1-score, Precision, and Recall reached 0.99854, 0.99709, 0.97321, 0.97396, 0.97344, and 0.97422, respectively. The proposed method outperforms the current major deep learning methods.

DPT-based seismic liquefaction triggering assessment in gravelly soils based on expanded case history dataset

The gravelly soil liquefaction caused by earthquake motions has been observed and reported in different regions in the world. However, few studies have been conducted to develop models for its prediction and risk assessment. Furthermore, the single available model was developed merely based on one case history earthquake. Thus, to bridge this gap in research, this study first created a comprehensive database of dynamic penetration test (DPT) using the documentation of eight global earthquakes, which includes a considerable variation of earthquake magnitudes, the frequencies of motions, epi-central distance, geometry type, deposit layering, gravelly soil depth, geometry type, and soil properties. Then, a deterministic model was developed to estimate the cyclic strength ratio (CRR) to estimate the safety factor (SF) against liquefaction triggering. Then, to account for uncertainties in the estimations of parameters and model prediction, Bayesian mapping function and logistic regression were applied to develop probabilistic models to classify liquefied and non-liquefied regions for liquefaction occurrence prediction. The maximizing likelihood function was used to calculate the model's parameters. The effect of the bias sampling factor was surveyed via supposing a range of variation for its value to discover its decent value. To validate the presented models, they were compared to the existed model. The results were studied via two issues, as a capability that shows the accuracy of the model's prediction and a risk issue via a recall index.

Research on Online Social Network Information Leakage-Tracking Algorithm Based on Deep Learning.

The rapid iteration of information technology makes the development of online social networks increasingly rapid, and its corresponding network scale is also increasingly large and complex. The corresponding algorithms to deal with social networks and their corresponding related problems are also increasing. The corresponding privacy protection algorithms such as encryption algorithm, access control strategy algorithm, and differential privacy protection algorithm have been studied and analyzed, but these algorithms do not completely solve the problem of privacy disclosure. Based on this, this article first searches and accurately filters the relevant information and content of online social networks based on the deep convolution neural network algorithm, so as to realize the perception and protection of users' safe content. For the corresponding graphics and data, this article introduces the compressed sensing technology to randomly disturb the corresponding graphics and data. At the level of tracking network information leakage algorithm, this article proposes a network information leakage-tracking algorithm based on digital fingerprint, which mainly uses relevant plug-ins to realize the unique identification processing of users, uses the uniqueness of digital fingerprint to realize the tracking processing of leakers, and formulates the corresponding coding scheme based on the social network topology, and at the same time, the network information leakage-tracking algorithm proposed in this article also has high efficiency in the corresponding digital coding efficiency and scalability. In order to verify the advantages of the online social network information leakage-tracking algorithm based on deep learning, this article compares it with the traditional algorithm. In the experimental part, this article mainly compares the accuracy index, recall index, and performance index. At the corresponding accuracy index level, it can be seen that the maximum improvement of the algorithm proposed in this article is about 10% compared with the traditional algorithm. At the corresponding recall index level, the proposed algorithm is about 5–8% higher than the traditional algorithm. Corresponding to the overall performance index, it improves the performance by about 50% compared with the traditional algorithm. The comparison results show that the proposed algorithm has higher accuracy and the corresponding source tracking is more accurate.