Contextual Features Research Articles

Progressively global–local fusion with explicit guidance for accurate and robust 3d hand pose reconstruction

Parametric and non-parametric methods are two commonly used strategies in current 3D hand pose reconstruction. Parametric methods predict low-dimensional parameters to fit a predefined hand model to the input image. Benefiting from the prior knowledge of hand models, parametric methods guarantee plausible hand poses, whereas the pose estimation accuracy is limited due to nonlinear regression and spatial information loss. Differently, non-parametric methods directly estimate the coordinates of keypoints or mesh vertices from the input image. The reconstructed 3D hand poses show high precision but may be less robust. In this paper, we integrate the advantages of two methods for accurate and robust hand pose reconstruction. Specifically, we disentangle the hand pose reconstruction into global modeling and local refinement, which is performed in a coarse-to-fine manner. Firstly, we utilize global features from the encoder to generate the initial estimation by a parametric method, which aims to provide the prior knowledge of the human hand for subsequent processes. Then, we gradually fuse multi-scale contextual features for local refinement by explicitly integrating global prior information and local visual features. In particular, we introduce a consecutive pixel-aligned feature retrieval module to extract fine-grained information from visual features, thereby achieving pixel-level alignment. Furthermore, we demonstrate that our method can be extended to weakly-supervised learning where only sparse pose annotations are needed, potentially alleviating the burden of meticulous mesh annotation. The effectiveness and robustness of our method are substantiated through both fully- and weakly-supervised experiments, demonstrating superior performance compared to state-of-the-art methods. We plan to release our code at https://github.com/Kun-Gao/P_GLFnet.

Pairwise Intent Graph Embedding Learning for Context-Aware Recommendation with Knowledge Graph

Different from the data sparsity that traditional recommendations suffer from, context-aware recommender systems (CARS) face specific sparsity challenges related to contextual features, i.e., feature sparsity and interaction sparsity. How knowledge graphs address these challenges remains under-discussed. To bridge this gap, in this paper, we first propose a novel pairwise intent graph (PIG) containing nodes of users, items, entities, and enhanced intents to integrate knowledge graphs into CARS efficiently. Enhanced intent nodes are generated through the specific fusion of relational sub-intent and contextual sub-intent, and they are derived from semantic information and contextual information, respectively. We develop a pairwise intent graph embedding learning (PING) framework based on it. Specifically, our PING uses a pairwise intent joint graph convolution module to obtain refined embedding of all the features, where each enhanced intent node acts as a hub to effectively propagate information among different features and between all the features and knowledge graphs. Then, a recommendation module with refined embeddings is used to replace the randomly initialized embeddings of downstream recommendation models to improve model performance. Extensive experiments on three public datasets and some real-world scenarios verify the effectiveness and compatibility of our PING.

An implicit aspect-based sentiment analysis method using supervised contrastive learning and knowledge embedding

Aspect-based sentiment analysis aims to analyze and understand people’s opinions from different aspects. Some comments do not contain explicit opinion words but still convey a clear human-perceived emotional orientation, which is known as implicit sentiment. Most previous research relies on contextual information from a text for implicit aspect-based sentiment analysis. However, little work has integrated external knowledge with contextual information. This paper proposes an implicit aspect-based sentiment analysis model combining supervised contrastive learning with knowledge-enhanced fine-tuning on BERT (BERT-SCL+KEFT). In the pre-training phase, the model utilizes supervised contrastive learning (SCL) on large-scale sentiment-annotated corpora to acquire sentiment knowledge. In the fine-tuning phase, the model uses a knowledge-enhanced fine-tuning (KEFT) method to capture explicit and implicit aspect-based sentiments. Specifically, the model utilizes knowledge embedding to embed external general knowledge information into textual entities by using knowledge graphs, enriching textual information. Finally, the model combines external knowledge and contextual features to predict the implicit sentiment in a text. The experimental results demonstrate that the proposed BERT-SCL+KEFT model outperforms other baselines on the general implicit sentiment analysis and implicit aspect-based sentiment analysis tasks. In addition, ablation experimental results show that the proposed BERT-SCL+KEFT model without the knowledge embedding module or supervised contrastive learning module significantly decreases performance, indicating the importance of these modules. All experiments validate that the proposed BERT-SCL+KEFT model effectively achieves implicit aspect-based sentiment classification.

Context propagation based influence maximization model for dynamic link prediction

Influence maximization (IM) in dynamic social networks is an optimization problem to analyze the changes in social networks for different periods. However, the existing IM methods ignore the context propagation of interaction behaviors among users. Hence, context-based IM in multiplex networks is proposed here. Initially, multiplex networks along with their contextual data are taken as input. Community detection is performed for the network using the Wilcoxon Hypothesized K-Means (WH-KMA) algorithm. From the detected communities, the homogeneous network is used for extracting network topological features, and the heterogeneous networks are used for influence path analysis based on which the node connections are weighted. Then, the influence-path-based features along with contextual features are extracted. These extracted features are given for the link prediction model using the Parametric Probability Theory-based Long Short-Term Memory (PPT-LSTM) model. Finally, from the network graph, the most influencing nodes are identified using the Linear Scaling based Clique (LS-Clique) detection algorithm. The experimental outcomes reveal that the proposed model achieves an enhanced performance.

Contextual feature fusion and refinement network for camouflaged object detection

Contextual feature fusion and refinement network for camouflaged object detection

Vivid London: Assessing the resilience of urban vibrancy during the COVID-19 pandemic using social media data

Since COVID-19, the focus on urban resilience has intensified, particularly on cities' ability to adapt and recover while maintaining essential functions and liveability; however, few studies have examined the resilience of urban vibrancy during such health crises. This study investigates urban vibrancy resilience in Inner London during the COVID-19 pandemic using multi-sourced social media data (geo-tagged Twitter and Flickr). We propose an analytical framework based on space-time permutation scan statistics (STPSS) to identify spatiotemporal urban areas of interest (ST-AOIs), examining their spatial, temporal, and contextual characteristics. Our findings show that central neighbourhoods with transport hubs, educational and healthcare facilities, eateries, and financial centres exhibit greater resilience. These areas adapt by shifting active periods in response to disruptions. Additionally, we assess the varying resilience capacities of different types of points of interest. This research provides actionable insights for urban planners and policymakers by demonstrating how identifying characteristics of robust urban vibrancy can contribute to the resilience of cities and communities, particularly under normal conditions after COVID-19. The findings offer concrete strategies for integrating social media data into urban planning processes, enabling more responsive and adaptive governance that meets the dynamic needs of urban populations.

Effects of the English language intervention as a foreign language for struggling elementary readers in South Korea: a meta-analysis

AbstractThis meta-analysis comprehensively examined the effects of English language interventions on struggling elementary school learners in South Korea learning English as a foreign language. A total of 22 studies of experimental and quasi-experimental studies were separately analyzed by the study design using a random-effects model, and the mean effect size was medium (g = 0.68, SE = 0.26, CI95 = 0.18, 1.18, p < 0.01) for the experimental studies while the mean effect size was large (g = 1.39, SE = 0.67, CI95 = 0.08, 2.69, p < 0.05) for the quasi-experimental studies. Due to the large homogeneity statistics, subgroup analyses were additionally conducted for possible moderator effects contributing to the mean. The subgroup analyses indicated large effects in the affective (e.g., student’s motivation) and achievement (e.g., reading comprehension) type of outcome measures and effective intervention components (e.g., shared reading). Other effective contextual characteristics including the total session numbers, intervention frequency, and classroom type are also discussed along with the study limitations and future directions for the struggling readers learning English as a foreign language.

Lightweight deep learning model for underwater waste segmentation based on sonar images

In recent years, the rapid accumulation of marine waste not only endangers the ecological environment but also causes seawater pollution. Traditional manual salvage methods often have low efficiency and pose safety risks to human operators, making automatic underwater waste recycling a mainstream approach. In this paper, we propose a lightweight multi-scale cross-level network for underwater waste segmentation based on sonar images that provides pixel-level location information and waste categories for autonomous underwater robots. In particular, we introduce hybrid perception and multi-scale attention modules to capture multi-scale contextual features and enhance high-level critical information, respectively. At the same time, we use sampling attention modules and cross-level interaction modules to achieve feature down-sampling and fuse detailed features and semantic features, respectively. Relevant experimental results indicate that our method outperforms other semantic segmentation models and achieves 74.66 % mIoU with only 0.68 M parameters. In particular, compared with the representative PIDNet Small model based on the convolutional neural network architecture, our method can improve the mIoU metric by 1.15 percentage points and can reduce model parameters by approximately 91 %. Compared with the representative SeaFormer T model based on the transformer architecture, our approach can improve the mIoU metric by 2.07 percentage points and can reduce model parameters by approximately 59 %. Our approach maintains a satisfactory balance between model parameters and segmentation performance. Our solution provides new insights into intelligent underwater waste recycling, which helps in promoting sustainable marine development.

Assessment of Eye Care Apps for Children and Adolescents Based on the Mobile App Rating Scale: Content Analysis and Quality Assessment.

In China, the current situation of myopia among children and adolescents is very serious. Prevention and control of myopia are inhibited by the lack of medical resources and the low awareness about eye care. Nevertheless, mobile apps provide an effective means to solve these problems. Since the health app market in China is still immature, it has become particularly important to conduct a study to assess the quality of eye-care apps to facilitate the development of better eye-care service strategies. This study aimed to evaluate the quality, functionality, medical evidence, and professional background of eye-care apps targeting children and adolescents in the Chinese app stores. A systematic search on iOS and Android app stores was performed to identify eye-care apps for children and adolescents. The general characteristics, development context, and functional features of the apps were described. Quality assessment of the apps was completed by 2 independent researchers using the Mobile App Rating Scale. This study included 29 apps, of which 17 (59%) were developed by commercial organizations and 12 (41%) had a design with relevant scientific basis. The main built-in functions of these apps include self-testing (18/29, 62%), eye exercises (16/29, 55%), and eye-care education (16/29, 55%). The mean overall quality of eye-care apps was 3.49 (SD 0.33), with a score ranging from 2.89 to 4.39. The overall Mobile App Rating Scale score exhibited a significant positive correlation with the subscale scores (r=0.81-0.91; P<.001). In addition, although most apps provided basic eye-care features, there are some deficiencies. For example, only a few apps were developed with the participation of medical organizations or professional ophthalmologists, and most of the apps were updated infrequently, failing to provide the latest eye-care information and technology in a timely manner. In general, the quality of eye-care apps for children and teenagers in Chinese app stores is good. These apps fulfill users' needs for eye-care services to a certain extent, but they still suffer from insufficient medical background, low user engagement, and untimely updates. In order to further improve the effectiveness of eye-care apps, cooperation with medical institutions and professional ophthalmologists should be strengthened to enhance the scientific and authoritative nature of the apps. At the same time, interactive features and regular updates should be added to enhance user participation and the continuity of the apps. This study provides a reference for future development or improvement of eye-care apps, which can help promote myopia prevention and control.

AMHFN: Aggregation Multi-Hierarchical Feature Network for Hyperspectral Image Classification

Deep learning methods like convolution neural networks (CNNs) and transformers are successfully applied in hyperspectral image (HSI) classification due to their ability to extract local contextual features and explore global dependencies, respectively. However, CNNs struggle in modeling long-term dependencies, and transformers may miss subtle spatial-spectral features. To address these challenges, this paper proposes an innovative hybrid HSI classification method aggregating hierarchical spatial-spectral features from a CNN and long pixel dependencies from a transformer. The proposed aggregation multi-hierarchical feature network (AMHFN) is designed to capture various hierarchical features and long dependencies from HSI, improving classification accuracy and efficiency. The proposed AMHFN consists of three key modules: (a) a Local-Pixel Embedding module (LPEM) for capturing prominent spatial-spectral features; (b) a Multi-Scale Convolutional Extraction (MSCE) module to capture multi-scale local spatial-spectral features and aggregate hierarchical local features; (c) a Multi-Scale Global Extraction (MSGE) module to explore multi-scale global dependencies and integrate multi-scale hierarchical global dependencies. Rigorous experiments on three public hyperspectral image (HSI) datasets demonstrated the superior performance of the proposed AMHFN method.

Socioformative process and educational benchmarking: implications for the quality of initial teacher training

In the Peruvian education system, permanent innovation in teacher training is achieved through the integration of emerging educational theories for teaching and learning, addressing the real needs of teachers in initial training within a context of continuous change. The objective of this research is to determine the level of influence of an educational model based on educational benchmarking and socio-formative processes on the improvement of the quality of initial teacher training. This research was quantitative, non-experimental, and cross-sectional. The population comprised 1,420 students of the Faculty of Education Sciences, with a study sample consisting of 142 students randomly chosen in a stratified manner. Data collection was carried out through a Likert-type questionnaire, followed by descriptive and inferential statistical analysis using Minitab software. The results indicate that 66.17% of the students agree with the practice of educational benchmarking, and 69.90% agree with the implementation of socio-educational activities. Additionally, 72.00% agree with the quality of professional training received at the university. The quality of vocational training is influenced by 77.39% through the praxis of socio-formative processes and educational benchmarking in university education. It is concluded that the practice of educational benchmarking and socio-training based on contextual characteristics significantly influences the quality of initial teacher training.

A social context-aware graph-based multimodal attentive learning framework for disaster content classification during emergencies

In times of crisis, the prompt and precise classification of disaster-related information shared on social media platforms is of paramount importance for effective disaster response and public safety. During such critical events, people utilize social media as a medium for communication, sharing multimodal textual and visual content. However, due to the substantial influx of unfiltered and diverse data, humanitarian organizations face challenges in effectively leveraging this information. Numerous methods have been proposed for classifying disaster-related content, but these methods lack modeling users’ credibility, emotional context, and social interaction information, which is crucial for classification. In this context, we propose a method, CrisisSpot, that leverages a Graph-based Neural Network to comprehend intricate relationships between textual and visual modalities and Social Context Features to incorporate user-centric and content-centric information. We also propose Inverted Dual Embedded Attention, which captures both harmonious and contrary patterns present in the data to harness complex interactions and facilitate richer insights in multimodal data. We have developed a multimodal disaster dataset, TSEqD (Turkey-Syria Earthquake Dataset), which is a large annotated dataset for a single disaster event containing 10,352 data samples. Through extensive experimentation, CrisisSpot has demonstrated significant improvements, achieving an average gain of 9.45% and 5.01% in F1-score compared to the state-of-the-art methods on the publicly available CrisisMMD dataset and TSEqD dataset, respectively.

Triple-Stream Commonsense Circulation Transformer Network for Image Captioning

Traditional image captioning methods only have a local perspective at the dataset level, allowing them to explore dispersed information within individual images. However, the lack of a global perspective prevents them from capturing common characteristics among similar images. To address the limitation, this paper introduces a novel Triple-stream Commonsense Circulating Transformer Network (TCCTN). It incorporates contextual stream into the encoder, combining enhanced channel stream and spatial stream for comprehensive feature learning. The proposed commonsense-aware contextual attention (CCA) module queries commonsense contextual features from the dataset, obtaining global contextual association information by projecting grid features into the contextual space. The pure semantic channel attention (PSCA) module leverages compressed spatial domain for channel pooling, focusing on attention weights of pure channel features to capture inherent semantic features. The region spatial attention (RSA) module enhances spatial concepts in semantic learning by incorporating region position information. Furthermore, leveraging the complementary differences among the three features, TCCTN introduces the mixture of experts strategy to enhance the unique discriminative ability of features and promote their integration in textual feature learning. Extensive experiments on the MS-COCO dataset demonstrate the effectiveness of contextual commonsense stream and the superior performance of TCCTN.

Factors influencing employers' support for employees with acquired brain injuries or mental illness to return to- and stay in work: A qualitative systematic review.

People with acquired brain injuries (ABIs) often experience residual limitations and co-morbid mental illnesses that restrict work participation. Employers are key in enabling successful return-to-work and job retention. This review aimed to explore employers' perspectives of factors influencing their support for people with ABIs and/or mental illness to return to- and stay in work. Review questions focused on barriers and facilitators to their support, and contextual characteristics present at the time. Five databases were searched from October 2010 until November 2023 for relevant qualitative studies published in English. Findings from included studies (N = 25) were synthesised using thematic synthesis. Included studies focused on employees with ABI or mental illness, rather than dually diagnosed ABI and mental illness. Employers' support was influenced by their awareness/knowledge of- and attitudes towards the employee's condition/illness; their skills and experience in supportive strategies; factors related to provision of work accommodations; and stakeholder influence. Similarities and differences in influential factors were observed across the ABI and mental illness literature. Contextual characteristics related to organisational characteristics, cultural taboo, and involvement of certain stakeholders. ABI survivors (with and without co-morbid mental illness) and their employers may benefit from specialist support and resources to guide them through the return-to-work process. Further research is needed to investigate employers' knowledge of ABI and mental illness and supportive strategies. Exploration of the influence of other stakeholders, socio-demographic characteristics, and contextual factors on employers' return-to-work and retention support for ABI survivors with co-morbid mental illness is warranted.

Intersectionality and Caregiving: The Exclusion Experience and Coping Resources of Immigrant Women Caring for a Family Member With Severe Mental Illness.

Intersectionality has become a central analytical framework in the study of exclusion and empowerment experiences among women from marginalized communities. However, the relevance of intersectionality to informal caregiving in mental healthcare has hardly been explored to date. The purpose of the current study is to examine the exclusion experiences and coping resources of immigrant women caring for a family member with a severe mental illness (SMI) through the lens of intersectionality theory. Semi-structured in-depth interviews were conducted with 26 informal female immigrant caregivers from the former Soviet Union residing in Israel. The interviews were analyzed using a qualitative content approach. The findings revealed that the participants experienced stigma and exclusion in several intersecting categories: economic marginalization of immigrant single mothers, ethnic and gender-based stigma of Russian-speaking women, gender-based domestic violence, and mental health stigma by professionals. The participants' coping resources included spirituality and religious faith, support groups, and social activism. The study provides insights into the burdens and rewards experienced by female immigrant caregivers of family members with SMI through the lens of intersectionality theory. Implications for adapting services to the contextual characteristics of female immigrant caregivers and minimizing intersectional stigma and inequities in informal healthcare are discussed.

Is Medium-term Exposure to Ambient Air Pollution a Risk Factor for Parkinson’s Disease?

ObjectivesThis study used Census data from Northern Ireland linked to administrative data on prescriptions between 2010 and 2016, to examine the association between ambient air pollution exposure and the risk of Parkinson’s Disease (PD) onset. It contributes to this literature by providing new evidence in a comparatively low pollution context, using rich and nationally representative cohort data with comprehensive information on Parkinson’s Disease medications dispensed over an extended period. ApproachWe estimated Cox Proportional Hazards models for the association between pollution exposure and PD onset as proxied by first receipt of PD medication, controlling extensively for potentially confounding factors at the neighbourhood, household, and individual levels. Estimates are presented in the form of hazard ratios for the effect of medium-term PM2.5 exposure on the risk of PD onset. ResultsThere was a clear unadjusted association between medium-term exposure to ambient pollution and the risk of developing PD, with those experiencing higher exposures being at greater risk. Once individual, household and neighbourhood confounders were adjusted for, however, we found no evidence supporting such associations. However, there was evidence of a positive association with both PM2.5 and NO2 exposure for sub-samples by age and education. ConclusionsThis study contributes to an emerging literature examining the association between ambient PM2.5 pollution and onset of PD. Despite finding an unconditional association, we find no evidence for an association once individual, family and contextual characteristics are controlled for, at least in the relatively low-pollution context of Northern Ireland.

Dynamic convolutional time series forecasting based on adaptive temporal bilateral filtering

Time series data typically contain complex dynamic patterns, which not only include linear trends and seasonal variations but also significant nonlinear changes and complex dependencies. Currently, feature extraction methods for time series data primarily employ mixed-mode extraction within the temporal domain, neglecting the effective extraction and analysis of nonlinear characteristics between different observation points and the interdependencies between variables. To address these issues, we designed an adaptive temporal bilateral filtering module that effectively preserves and highlights the nonlinear features and patterns in time series while filtering out noise and redundant information. We also designed a nonlinear feature adaptive extraction module that integrates a gating mechanism and deformable convolutions. This design allows the model to adaptively adjust the shape of the convolutional kernels according to different time steps and conditions. This enables accurate capture and extraction of nonlinear features between various time observations and dependencies between different variables. Additionally, we use stacked convolutional layers to extract local contextual features, addressing fluctuations in local features caused by changes in data distribution in real-world scenarios. In summary, we propose DCNet, a dynamic convolutional network based on an adaptive temporal bilateral filter, and evaluated its performance on twelve real-world datasets. The results indicate that DCNet consistently achieves state-of-the-art performance in both short-term and long-term forecasting tasks, with favorable runtime efficiency.

Using syntax and semantics to acquire subjective adjective meanings

ABSTRACT Adjectives like good, nice, pretty, and big are among the most frequent adjectives heard by children in the ambient language, and among the first adjectives produced by children. While these gradable adjectives are ostensibly short and sweet, their abstract semantic representations encode reference to features of the discourse context, including speaker subjectivity, which are surprisingly complex, and which also give rise to specific subclasses. How, then, do children acquire these fine-grained adjectival meanings? In this research, we hypothesize that as with verbs, the linguistic environments in which these adjectives appear provide surface-level distributional cues that narrow the space of meaning. In a proof-of-concept experiment with adults following in the footsteps of the Human Simulation Paradigm, we provide empirical evidence that surface-level linguistic patterns support the acquisition of different subclasses of subjective adjectives. Adults were presented with scripted dialogues featuring novel words in controlled linguistic environments manipulating grammatical subjects, adverbial modifiers, standard phrases, syntactic complements, and evidence of faultless disagreement among speakers. Participants were then asked to guess the word meaning (either incrementally after each ‘frame’ or after multiple frames) and rate their confidence. Results, as measured by the percentage of category-consistent, frame-compliant guesses, paired with an increase in confidence levels, indicate a sensitivity to this surface-level information, thereby highlighting the role of the linguistic environment in supporting the acquisition of adjective meaning.

Aspect-based sentiment analysis via bidirectional variant spiking neural P systems

In recent years, aspect-based sentiment classification has predominantly relied on artificial neural networks, achieving notable success. However, effectively capturing the contextual semantic relationships between aspects and content has remained challenging. To address this issue, we propose an innovative approach for aspect-based sentiment analysis. Our method uses a Bidirectional Variant Spiking Neural P System (Bi-VSNP) as a context encoding module, integrated with a two-channel mechanism to extract context and aspect features. Furthermore, Our dual attention mechanism leverages the relationship between aspects and context, allowing the model to focus on key sentiment information. We rigorously evaluated our method on publicly available datasets, demonstrating its superior performance over most models based on artificial neural networks as context extraction. Compared to other models utilizing SNP methods, our approach achieves performance improvements of 1.72%, 1.27%, and 1.05% on the Laptop, Restaurant, and Twitter datasets, respectively. These results underscore the efficacy of our approach in aspect-based sentiment classification.

Contextual visual and motion salient fusion framework for action recognition in dark environments

Infrared (IR) human action recognition (AR) exhibits resilience against shifting illumination conditions, changes in appearance, and shadows. It has valuable applications in numerous areas of future sustainable and smart cities including robotics, intelligent systems, security, and transportation. However, current IR-based recognition approaches predominantly concentrate on spatial or local temporal information and often overlook the potential value of global temporal patterns. This oversight can lead to incomplete representations of body part movements and prevent accurate optimization of a network. Therefore, a contextual-motion coalescence network (CMCNet) is proposed that operates in a streamlined and end-to-end manner for robust action representation in darkness in a near-infrared (NIR) setting. Initially, data are preprocessed to separate foreground, normalized, and resized. The framework employs two parallel modules: the contextual visual features learning module (CVFLM) for local feature extraction, and the temporal optical flow learning module (TOFLM) for acquiring motion dynamics. These modules focus on action-relevant regions used shift window-based operations to ensure accurate interpretation of motion information. The coalescence block harmoniously integrates the contextual and motion features within a unified framework. Finally, the temporal decoder module discriminatively identifies the boundaries of the action sequence. This sequence of steps ensures the synergistic optimization of both CVFLM and TOFLM and thorough competent feature extraction for precise AR. Evaluations of CMCNet are carried out on publicly available datasets, InfAR and NTURGB-D, where superior performance is achieved. Our model yields the highest average precision of 89% and 85% on these datasets, respectively, representing an improvement of 2.25% (on InfAR) compared to conventional methods operating at spatial and optical flow levels which underscores its efficacy.