Search results

Although numerous signal modalities are available for emotion recognition, audio and visual modalities are the most common and predominant forms for human beings to express their emotional states in daily communication. Therefore, how to achieve automatic and accurate audiovisual emotion recognition is significantly important for developing engaging and empathetic human–computer interaction environment. However, two major challenges exist in the field of audiovisual emotion recognition: (1) how to effectively capture representations of each single modality and eliminate redundant features and (2) how to efficiently integrate information from these two modalities to generate discriminative representations.

A novel key-frame extraction-based attention fusion network (KE-AFN) is proposed for audiovisual emotion recognition. KE-AFN attempts to integrate key-frame extraction with multimodal interaction and fusion to enhance audiovisual representations and reduce redundant computation, filling the research gaps of existing approaches. Specifically, the local maximum–based content analysis is designed to extract key-frames from videos for the purpose of eliminating data redundancy. Two modules, including “Multi-head Attention-based Intra-modality Interaction Module” and “Multi-head Attention-based Cross-modality Interaction Module”, are proposed to mine and capture intra- and cross-modality interactions for further reducing data redundancy and producing more powerful multimodal representations.

Extensive experiments on two benchmark datasets (i.e. RAVDESS and CMU-MOSEI) demonstrate the effectiveness and rationality of KE-AFN. Specifically, (1) KE-AFN is superior to state-of-the-art baselines for audiovisual emotion recognition. (2) Exploring the supplementary and complementary information of different modalities can provide more emotional clues for better emotion recognition. (3) The proposed key-frame extraction strategy can enhance the performance by more than 2.79 per cent on accuracy. (4) Both exploring intra- and cross-modality interactions and employing attention-based audiovisual fusion can lead to better prediction performance.

The proposed KE-AFN can support the development of engaging and empathetic human–computer interaction environment.

Software defect prediction (SDP) is a critical aspect of software quality assurance, aiming to identify and manage potential defects in software systems. In this paper, we have proposed a novel hybrid approach that combines Gray Wolf Optimization with Feature Selection (GWOFS) and multilayer perceptron (MLP) for SDP. The GWOFS-MLP hybrid model is designed to optimize feature selection, ultimately enhancing the accuracy and efficiency of SDP. Gray Wolf Optimization, inspired by the social hierarchy and hunting behavior of gray wolves, is employed to select a subset of relevant features from an extensive pool of potential predictors. This study investigates the key challenges that traditional SDP approaches encounter and proposes promising solutions to overcome time complexity and the curse of the dimensionality reduction problem.

The integration of GWOFS and MLP results in a robust hybrid model that can adapt to diverse software datasets. This feature selection process harnesses the cooperative hunting behavior of wolves, allowing for the exploration of critical feature combinations. The selected features are then fed into an MLP, a powerful artificial neural network (ANN) known for its capability to learn intricate patterns within software metrics. MLP serves as the predictive engine, utilizing the curated feature set to model and classify software defects accurately.

The performance evaluation of the GWOFS-MLP hybrid model on a real-world software defect dataset demonstrates its effectiveness. The model achieves a remarkable training accuracy of 97.69% and a testing accuracy of 97.99%. Additionally, the receiver operating characteristic area under the curve (ROC-AUC) score of 0.89 highlights the model’s ability to discriminate between defective and defect-free software components.

Experimental implementations using machine learning-based techniques with feature reduction are conducted to validate the proposed solutions. The goal is to enhance SDP’s accuracy, relevance and efficiency, ultimately improving software quality assurance processes. The confusion matrix further illustrates the model’s performance, with only a small number of false positives and false negatives.

The study aims to propose an intelligent real-time traffic model to address the traffic congestion problem. The proposed model assists the urban population in their everyday lives by assessing the probability of road accidents and accurate traffic information prediction. It also helps in reducing overall carbon dioxide emissions in the environment and assists the urban population in their everyday lives by increasing overall transportation quality.

This study offered a real-time traffic model based on the analysis of numerous sensor data. Real-time traffic prediction systems can identify and visualize current traffic conditions on a particular lane. The proposed model incorporated data from road sensors as well as a variety of other sources. It is difficult to capture and process large amounts of sensor data in real time. Sensor data is consumed by streaming analytics platforms that use big data technologies, which is then processed using a range of deep learning and machine learning techniques.

The study provided in this paper would fill a gap in the data analytics sector by delivering a more accurate and trustworthy model that uses internet of things sensor data and other data sources. This method can also assist organizations such as transit agencies and public safety departments in making strategic decisions by incorporating it into their platforms.

The model has a big flaw in that it makes predictions for the period following January 2020 that are not particularly accurate. This, however, is not a flaw in the model; rather, it is a flaw in Covid-19, the global epidemic. The global pandemic has impacted the traffic scenario, resulting in erratic data for the period after February 2020. However, once the circumstance returns to normal, the authors are confident in their model’s ability to produce accurate forecasts.

To help users choose when to go, this study intended to pinpoint the causes of traffic congestion on the highways in the Bay Area as well as forecast real-time traffic speeds. To determine the best attributes that influence traffic speed in this study, the authors obtained data from the Caltrans performance measurement system (PeMS), reviewed it and used multiple models. The authors developed a model that can forecast traffic speed while accounting for outside variables like weather and incident data, with decent accuracy and generalizability. To assist users in determining traffic congestion at a certain location on a specific day, the forecast method uses a graphical user interface. This user interface has been designed to be readily expanded in the future as the project’s scope and usefulness increase. The authors’ Web-based traffic speed prediction platform is useful for both municipal planners and individual travellers. The authors were able to get excellent results by using five years of data (2015–2019) to train the models and forecast outcomes for 2020 data. The authors’ algorithm produced highly accurate predictions when tested using data from January 2020. The benefits of this model include accurate traffic speed forecasts for California’s four main freeways (Freeway 101, I-680, 880 and 280) for a specific place on a certain date. The scalable model performs better than the vast majority of earlier models created by other scholars in the field. The government would benefit from better planning and execution of new transportation projects if this programme were to be extended across the entire state of California. This initiative could be expanded to include the full state of California, assisting the government in better planning and implementing new transportation projects.

To estimate traffic congestion, the proposed model takes into account a variety of data sources, including weather and incident data. According to traffic congestion statistics, “bottlenecks” account for 40% of traffic congestion, “traffic incidents” account for 25% and “work zones” account for 10% (Traffic Congestion Statistics). As a result, incident data must be considered for analysis. The study uses traffic, weather and event data from the previous five years to estimate traffic congestion in any given area. As a result, the results predicted by the proposed model would be more accurate, and commuters who need to schedule ahead of time for work would benefit greatly.

The proposed work allows the user to choose the optimum time and mode of transportation for them. The underlying idea behind this model is that if a car spends more time on the road, it will cause traffic congestion. The proposed system encourages users to arrive at their location in a short period of time. Congestion is an indicator that public transportation needs to be expanded. The optimum route is compared to other kinds of public transit using this methodology (Greenfield, 2014). If the commute time is comparable to that of private car transportation during peak hours, consumers should take public transportation.

While the Chinese securities market is booming, the phenomenon of listed companies falling into financial distress is also emerging, which affects the operation and development of enterprises and also jeopardizes the interests of investors. Therefore, it is important to understand how to accurately and reasonably predict the financial distress of enterprises.

In the present study, ensemble feature selection (EFS) and improved stacking were used for financial distress prediction (FDP). Mutual information, analysis of variance (ANOVA), random forest (RF), genetic algorithms, and recursive feature elimination (RFE) were chosen for EFS to select features. Since there may be missing information when feeding the results of the base learner directly into the meta-learner, the features with high importance were fed into the meta-learner together. A screening layer was added to select the meta-learner with better performance. Finally, Optima hyperparameters were used for parameter tuning by the learners.

An empirical study was conducted with a sample of A-share listed companies in China. The F1-score of the model constructed using the features screened by EFS reached 84.55%, representing an improvement of 4.37% compared to the original features. To verify the effectiveness of improved stacking, benchmark model comparison experiments were conducted. Compared to the original stacking model, the accuracy of the improved stacking model was improved by 0.44%, and the F1-score was improved by 0.51%. In addition, the improved stacking model had the highest area under the curve (AUC) value (0.905) among all the compared models.

Compared to previous models, the proposed FDP model has better performance, thus bridging the research gap of feature selection. The present study provides new ideas for stacking improvement research and a reference for subsequent research in this field.

This study aims to analyse the systemic risk emitted by all publicly listed commercial banks in a key emerging economy, India.

The study makes use of the Tobias and Brunnermeier (2016) estimator to quantify the systemic risk (ΔCoVaR) that banks contribute to the system. The methodology addresses a classification problem based on the probability that a particular bank will emit high systemic risk or moderate systemic risk. The study applies machine learning models such as logistic regression, random forest (RF), neural networks and gradient boosting machine (GBM) and addresses the issue of imbalanced data sets to investigate bank’s balance sheet features and bank’s stock features which may potentially determine the factors of systemic risk emission.

The study reports that across various performance matrices, the authors find that two specifications are preferred: RF and GBM. The study identifies lag of the estimator of systemic risk, stock beta, stock volatility and return on equity as important features to explain emission of systemic risk.

The findings will help banks and regulators with the key features that can be used to formulate the policy decisions.

This study contributes to the existing literature by suggesting classification algorithms that can be used to model the probability of systemic risk emission in a classification problem setting. Further, the study identifies the features responsible for the likelihood of systemic risk.

The development of information technologies has led to a considerable transformation in human resource management from conventional or commonly known as personnel management to modern one. Data mining technology, which has been widely used in several applications, including those that function on the web, includes clustering algorithms as a key component. Web intelligence is a recent academic field that calls for sophisticated analytics and machine learning techniques to facilitate information discovery, particularly on the web. Human resource data gathered from the web are typically enormous, highly complex, dynamic, and unstructured. Traditional clustering methods need to be upgraded because they are ineffective. Standard clustering algorithms are enhanced and expanded with optimization capabilities to address this difficulty by swarm intelligence, a subset of nature-inspired computing. We collect the initial raw human resource data and preprocess the data wherein data cleaning, data normalization, and data integration takes place. The proposed K-C-means-data driven cuckoo bat optimization algorithm (KCM-DCBOA) is used for clustering of the human resource data. The feature extraction is done using principal component analysis (PCA) and the classification of human resource data is done using support vector machine (SVM). Other approaches from the literature were contrasted with the suggested approach. According to the experimental findings, the suggested technique has extremely promising features in terms of the quality of clustering and execution time.

This study aims to identify an automated machine learning algorithm with high accuracy that sport practitioners can use to identify the specific factors for predicting Major League Baseball (MLB) attendance. Furthermore, by predicting spectators for each league (American League and National League) and division in MLB, the authors will identify the specific factors that increase accuracy, discuss them and provide implications for marketing strategies for academics and practitioners in sport.

This study used six years of daily MLB game data (2014–2019). All data were collected as predictors, such as game performance, weather and unemployment rate. Also, the attendance rate was obtained as an observation variable. The Random Forest, Lasso regression models and XGBoost were used to build the prediction model, and the analysis was conducted using Python 3.7.

The RMSE value was 0.14, and the R2 was 0.62 as a consequence of fine-tuning the tuning parameters of the XGBoost model, which had the best performance in forecasting the attendance rate. The most influential variables in the model are “Rank” of 0.247 and “Day of the week”, “Home team” and “Day/Night game” were shown as influential variables in order. The result was shown that the “Unemployment rate”, as a macroeconomic factor, has a value of 0.06 and weather factors were a total value of 0.147.

This research highlights unemployment rate as a determinant affecting MLB game attendance rates. Beyond contextual elements such as climate, the findings of this study underscore the significance of economic factors, particularly unemployment rates, necessitating further investigation into these factors to gain a more comprehensive understanding of game attendance.

This study examines how dynamic capabilities (DCs) impact global value chain (GVC) upgrading and assesses the mediating role of innovation.

The study uses a quantitative research method. The data are collected using an online questionnaire administered to respondents working in Chinese automobile manufacturers in China and Laos. The data are analyzed using structural equation modeling (SEM) and related software.

The results show that DCs and innovation capabilities (ICs) positively affect GVC upgrading and that ICs plays a mediating role between DC and GVC upgrading. Dynamic capabilities evolution (CE) mediates the relationship between DCs, ICs and GVC upgrading. Finally, differences exist in the effects of the three dimensions of DCs on ICs and GVC upgrading.

Focusing on the absorption and transformation of knowledge, enterprises could experience a clear enhancement of IC and CE and be more likely to obtain higher marginal returns. The study provides insights for emerging market firms to gain higher added value in internationalization.

This study demonstrates that different dimensions of DCs have different effects on GVC upgrading. In terms of theory, the impact of IC is considered in terms of the mediating effect of CE on IC. Differences are highlighted concerning the impact of learning capability, integrating and coordinating capability and sensing capability on the mediated relationships.

Nowadays, artificial intelligence (AI) technology has demonstrated extensive applications in the field of art design. Attribute editing is an important means to realize clothing style and color design via computer language, which aims to edit and control the garment image based on the specified target attributes while preserving other details from the original image. The current image attribute editing model often generates images containing missing or redundant attributes. To address the problem, this paper aims for a novel design method utilizing the Fashion-attribute generative adversarial network (AttGAN) model was proposed for image attribute editing specifically tailored to women’s blouses.

The proposed design method primarily focuses on optimizing the feature extraction network and loss function. To enhance the feature extraction capability of the model, an increase in the number of layers in the feature extraction network was implemented, and the structure similarity index measure (SSIM) loss function was employed to ensure the independent attributes of the original image were consistent. The characteristic-preserving virtual try-on network (CP_VTON) dataset was used for train-ing to enable the editing of sleeve length and color specifically for women’s blouse.

The experimental results demonstrate that the optimization model’s generated outputs have significantly reduced problems related to missing attributes or visual redundancy. Through a comparative analysis of the numerical changes in the SSIM and peak signal-to-noise ratio (PSNR) before and after the model refinement, it was observed that the improved SSIM increased substantially by 27.4%, and the PSNR increased by 2.8%, serving as empirical evidence of the effectiveness of incorporating the SSIM loss function.

The proposed algorithm provides a promising tool for precise image editing of women’s blouses based on the GAN. This introduces a new approach to eliminate semantic expression errors in image editing, thereby contributing to the development of AI in clothing design.

Limited by the types of sensors, the state information available for musculoskeletal robots with highly redundant, nonlinear muscles is often incomplete, which makes the control face a bottleneck problem. The aim of this paper is to design a method to improve the motion performance of musculoskeletal robots in partially observable scenarios, and to leverage the ontology knowledge to enhance the algorithm’s adaptability to musculoskeletal robots that have undergone changes.

A memory and attention-based reinforcement learning method is proposed for musculoskeletal robots with prior knowledge of muscle synergies. First, to deal with partially observed states available to musculoskeletal robots, a memory and attention-based network architecture is proposed for inferring more sufficient and intrinsic states. Second, inspired by muscle synergy hypothesis in neuroscience, prior knowledge of a musculoskeletal robot’s muscle synergies is embedded in network structure and reward shaping.

Based on systematic validation, it is found that the proposed method demonstrates superiority over the traditional twin delayed deep deterministic policy gradients (TD3) algorithm. A musculoskeletal robot with highly redundant, nonlinear muscles is adopted to implement goal-directed tasks. In the case of 21-dimensional states, the learning efficiency and accuracy are significantly improved compared with the traditional TD3 algorithm; in the case of 13-dimensional states without velocities and information from the end effector, the traditional TD3 is unable to complete the reaching tasks, while the proposed method breaks through this bottleneck problem.

In this paper, a novel memory and attention-based reinforcement learning method with prior knowledge of muscle synergies is proposed for musculoskeletal robots to deal with partially observable scenarios. Compared with the existing methods, the proposed method effectively improves the performance. Furthermore, this paper promotes the fusion of neuroscience and robotics.