Search results

The purpose of this study is to explore the potential of trainable activation functions to enhance the performance of deep neural networks, specifically ResNet architectures, in the task of image classification. By introducing activation functions that adapt during training, the authors aim to determine whether such flexibility can lead to improved learning outcomes and generalization capabilities compared to static activation functions like ReLU. This research seeks to provide insights into how dynamic nonlinearities might influence deep learning models' efficiency and accuracy in handling complex image data sets.

This research integrates three novel trainable activation functions – CosLU, DELU and ReLUN – into various ResNet-n architectures, where “n” denotes the number of convolutional layers. Using CIFAR-10 and CIFAR-100 data sets, the authors conducted a comparative study to assess the impact of these functions on image classification accuracy. The approach included modifying the traditional ResNet models by replacing their static activation functions with the trainable variants, allowing for dynamic adaptation during training. The performance was evaluated based on accuracy metrics and loss profiles across different network depths.

The findings indicate that trainable activation functions, particularly CosLU, can significantly enhance the performance of deep learning models, outperforming the traditional ReLU in deeper network configurations on the CIFAR-10 data set. CosLU showed the highest improvement in accuracy, whereas DELU and ReLUN offered varying levels of performance enhancements. These functions also demonstrated potential in reducing overfitting and improving model generalization across more complex data sets like CIFAR-100, suggesting that the adaptability of activation functions plays a crucial role in the training dynamics of deep neural networks.

This study contributes to the field of deep learning by introducing and evaluating the impact of three novel trainable activation functions within widely used ResNet architectures. Unlike previous works that primarily focused on static activation functions, this research demonstrates that incorporating trainable nonlinearities can lead to significant improvements in model performance and adaptability. The introduction of CosLU, DELU and ReLUN provides a new pathway for enhancing the flexibility and efficiency of neural networks, potentially setting a new standard for future deep learning applications in image classification and beyond.

This study aims to promote the anti-disturbance and tracking accuracy performance of the servo systems, in which a modified active disturbance rejection control (MADRC) scheme is proposed.

An adaptive radial basis function (ARBF) neural network is utilized to estimate and compensate dominant friction torque disturbance, and a parallel high-gain extended state observer (PHESO) is employed to further compensate residual and other uncertain disturbances. This parallel compensation structure reduces the burden of single ESO and improves the response speed of permanent magnet synchronous motor (PMSM) to hybrid disturbances. Moreover, the sliding mode control (SMC) rate is introduced to design an adaptive update law of ARBF.

Simulation and experimental results show that as compared to conventional ADRC and SMC algorithms, the position tracking error is only 2.3% and the average estimation error of the total disturbances is only 1.4% in the proposed MADRC algorithm.

The disturbance parallel estimation structure proposed in MADRC algorithm is proved to significantly improve the performance of anti-disturbance and tracking accuracy.

The research attempts to estimate how the use of WeChat, the most popular mobile social networking application in contemporary China, affects rural household income.

Our materials are 4,552 rural samples from the Chinese General Social Survey, and a treatment effect (TE) model is employed to address the endogeneity of WeChat usage.

The results prove that WeChat usage has a statistically significant and positive correlation with rural household income. This conclusion remains robust after using alternative variables to replace the explanatory and dependent variables. Our research provides two channels through which WeChat usage boosts rural household income, namely, it can promote their off-farm employment and participation in investment activities.

Theoretically, the study provides several micro-evidences for understanding the impact of mobile social networks on rural household welfare. Further, our findings may shed light on the importance of digital technology applications in rural poverty alleviation for developing countries.

Green hospitality studies have not adequately focused on the diffusion of eco-innovative hotels amongst visitors. This study aims to fill this gap by identifying green hotel attributes that influence visitors’ adoption of eco-friendly hotel and their intentions to partake in green initiatives.

The paper uses a mixed-method approach to explore the drivers of customers’ green hotel adoption and consumption. In the qualitative phase, data were collected via 20 open-ended interviews and analyzed to derive a measurement scale. The scale was then tested through a survey comprising 500 respondents using structural equation modelling.

The study results elucidate how guests’ visit intentions and green consumption behavior is built through their perception of newness and uniqueness of eco-innovative attributes. Findings shed light on how green hotel’s sustainable communication and corporate social responsibility outreach efforts positively influence guest visit intentions.

Study results reveal perceived eco-innovativeness as an important antecedent of visit intentions. Based on guest’s preferences, green hotels striving to increase its visitors’ base could begin by expanding their eco-innovative attributes.

Contrasting previous studies that have exclusively used the theory of planned behavior constructs, this study argues that diffusion of innovation constructs also offer valuable insights into guests’ visit intentions. While existing studies have covered limited number of eco-innovative attributes, this study adds to the literature by presenting a comprehensive set of attributes including trustworthiness of communication and observability of its social impacts.

Devices for step estimation are body-worn devices used to compute steps taken and/or distance covered by the user. Even though textiles or clothing are foremost to come to mind in terms of articles meant to be worn, their prominence among devices and systems meant for cadence is overshadowed by electronic products such as accelerometers, wristbands and smart phones. Athletes and sports enthusiasts using knee sleeves should be able to track their performances and monitor workout progress without the need to carry other devices with no direct sport utility, such as wristbands and wearable accelerometers. The purpose of this study thus is to contribute to the broad area of wearable devices for cadence application by developing a cheap but effective and efficient stride measurement system based on a knee sleeve.

A textile strain sensor is designed by weft knitting silver-plated nylon yarn together with nylon DTY and covered elastic yarn using a 1 × 1 rib structure. The area occupied by the silver-plated yarn within the structure served as the strain sensor. It worked such that, upon being subjected to stress, the electrical resistance of the sensor increases and in turn, is restored when the stress is removed. The strip with the sensor is knitted separately and subsequently sewn to the knee sleeve. The knee sleeve is then connected to a custom-made signal acquisition and processing system. A volunteer was employed for a wearer trial.

Experimental results establish that the number of strides taken by the wearer can easily be correlated to the knee flexion and extension cycles of the wearer. The number of peaks computed by the signal acquisition and processing system is therefore counted to represent stride per minute. Therefore, the sensor is able to effectively count the number of strides taken by the user per minute. The coefficient of variation of over-ground test results yielded 0.03%, and stair climbing also obtained 0.14%, an indication of very high sensor repeatability.

The study was conducted using limited number of volunteers for the wearer trials.

By embedding textile piezoresistive sensors in some specific garments and or accessories, physical activity such as gait and its related data can be effectively measured.

To the best of our knowledge, this is the first application of piezoresistive sensing in the knee sleeve for stride estimation. Also, this study establishes that it is possible to attach (sew) already-knit textile strain sensors to apparel to effectuate smart functionality.

Civil service workers are valuable resources for any nation and play a crucial role in driving their country’s economic development. Per the supervisor, this research examines the impact of mindfulness, proactive personality, and career competencies on employee job performance. The study also analyzes the effects of career adaptability and identity on this aspect.

To test the model of this study, questionnaires were administered to a sample of 500 civil service employees whose career-based knowledge and skills were measured in various cities in the province of Punjab, Pakistan.

Mindfulness and career competencies significantly impact supervisor-rated task performance, whereas a proactive personality does not substantially relate to supervisor-rated task performance. Research indicated that the two hypotheses about mediation were accepted. However, career adaptability does not play a significant role in the link between mindfulness and how well a supervisor rates task performance. Regarding moderation, career identity did not significantly moderate the relation between proactive personality and supervisor-rated task performance. However, the other two moderate hypotheses have been proven to be significant.

The findings offer compelling support for career construction theory (CCT) in this study area by analyzing the connections related to career adaptability and identity within the framework. In the future, researchers can build on this model by adding theories like conservation of resources (COR), looking into possible moderators that might change specific pathways in this network of relationships and using longitudinal designs to find stronger causal relationships.

Considering the evolving workplace due to the COVID-19 pandemic, the study offers fresh perspectives on the post-COVID situation, understanding and integrating various variables. For future studies, more variables can be explored in this model with the expansion of sample size and change of context.

This study examines the non-linear effect of board independence on the investment efficiency of listed firms worldwide. This study further tests whether the COVID-19 pandemic, industry competition and economic development influence the relationship between board independence and investment efficiency.

The data are retrieved from the Thomson Reuters (Refinitiv) database and include international data from 33 countries, comprising 21,363 firm-year observations. The authors' regression analyses include firm-specific variables as controls that may impact investment efficiency. The authors also perform various robustness tests including, alternative measures of investment efficiency, weighted least squares regression, quantile regression and endogeneity issues.

The results reveal a non-linear relationship between board independence and investment efficiency. Specifically, the relationship follows a U-shaped pattern, indicating that the negative impact of board independence on investment efficiency becomes positive after it reaches its optimal point, thus supporting optimal board structure theory. Interestingly, the authors find no significant evidence of board independence’s effect on investment efficiency during the pandemic. In contrast, the relationship between board independence and investment efficiency is significant only during the non-pandemic period. Furthermore, the authors discover evidence of a U-shaped relationship in both emerging and developed markets, as well as in industries with high and low competition.

The authors' study discovers new evidence on the non-linear impact of board independence on investment efficiency, which has not been explored previously in existing research.

This study has practical implications for investors by emphasising the importance of corporate governance and the appointment of independent directors. Investors should consider the findings of this study when making decisions related to corporate governance, as they can impact a firm's investment efficiency.

Despite a considerable body of literature exploring the link between corporate governance and investment effectiveness, there is a dearth of research on the non-linear effects of board independence. Furthermore, the effects of the COVID-19 pandemic, industry competition and economic development remain unexplored.

Most research on sustainable tourism has been devoted to understanding the determinants of tourists' sustainable behavior on a unidimensional construct, overlooking the importance of behavioral costs in sustainable travel behavior. To shed light on this issue, this study aims to quantitatively differentiate sustainable travel behaviors based on behavioral costs and to examine the impact of psychological factors on both low-cost and high-cost sustainable travel behaviors.

A survey of 470 tourists used Rasch analysis to measure the behavioral costs associated with sustainable travel behavior and partial least squares structural equation modeling (PLS-SEM) to test hypotheses.

The results indicate that the value-identity-personal norm model explains more variance in low-cost sustainable travel behaviors than in high-cost sustainable travel behaviors. This supports the central tenet of the low-cost hypothesis and also suggests that values and self-identity factors have a stronger influence on low-cost sustainable travel behavior. However, personal norms have a stronger influence on high-cost behaviors.

This research highlights the importance for tourism and destination managers to distinguish between different categories of sustainable travel behavior and to analyze their determinants separately. This allows for the development of tailored messages for specific groups of tourists based on the psychological drivers of sustainable travel behavior.

This study provides insights into the determinants of sustainable travel behaviors with different behavioral costs and highlights the importance of analyzing different categories of behaviors separately.