Accelerated Federated Learning Using Self-Adapting Bat Algorithm DOI Creative Commons

Jie Wang,

Chaochao Sun, Peng Yuan

и другие.

Research Square (Research Square), Год журнала: 2024, Номер unknown

Опубликована: Авг. 12, 2024

Abstract Federated learning (FL) is an advanced distributed machine (ML) framework designed to address issues related data silos and privacy. A significant challenge in FL the non-independent identically (Non-IID) nature of client data, resulting slow convergence rate low prediction accuracy for model. To tackle these issues, we propose a scheme based on bat algorithm (FedBat), leveraging echolocation mechanism bats effectively balance global local search capabilities optimizing model weight updates through dynamic adjustments strategy. FedBat also allows adaptive parameter across various datasets. mitigate drift issue, extend by using Jensen-Shannon(JS) divergence quantify difference between models. Clients decide whether upload their models this difference, aiming enhance model's generalization capability minimize communication overhead. Experimental results demonstrate that converges 5 times faster enhances test more than 40% compared FedAvg. The extended mitigates decrease performance reduces costs around 20%. Comparing FedPSO, FedGwo, FedProx shows demonstrates superior terms accuracy. We derive formula expected FedBat, analyze impact parameters performance, establish upper bound evaluate its divergence.

Язык: Английский

Influence of ChatGPT in professional communication – moderating role of perceived innovativeness DOI
Smriti Mathur, V Anand, Durgansh Sharma

и другие.

International Journal of Information and Learning Technology, Год журнала: 2025, Номер unknown

Опубликована: Янв. 1, 2025

Purpose ChatGPT, a cutting-edge language model, stands as an unparalleled, unmatched conversational ally, showcasing novel versatility and intelligence in its responses. This research delves into the incorporation of powerful generative AI tool, professional communication. study utilizes information system success model (ISSM) to examine role ChatGPTs strengthening quality (IQ), (SQ) service (SEQ) for improving customer usage intention (UI) satisfaction (SAT). The also investigates moderating impact perceived innovativeness between these relationships. Design/methodology/approach collected data from sample 400 customers through online survey validated hypothesized relationships using structural equation modelling (SEM). Process Macros 4.1 SPSS 22.0 is used test innovation IQ, SQ SEQ UI SAT. Findings results SEM analysis indicate that all positively support use ChatGPT communication with result establishes moderates relationship Originality/value offers contributions literature body knowledge by establishing Further, this proposes 2*2 matrix segment SAT users varying degrees innovativeness.

Язык: Английский

Процитировано

1
The Emerging Phenomenon of Shopstreaming: Gaining a More Nuanced Understanding of the Factors Which Drive It DOI Creative Commons
Ibrahim Mutambik

Journal of theoretical and applied electronic commerce research, Год журнала: 2024, Номер 19(3), С. 2522 - 2542

Опубликована: Сен. 23, 2024

Over the past decade, concept and practice of shopstreaming (also known as livestream shopping) have grown significantly within e-business world, it integrates live streaming technology with e-commerce. However, relationship between perceived benefits this shopping mode intention to use is not fully understood. This research seeks enhance current understanding by studying association in context fashion personal care (FPC) goods. Uniquely, study bases its core model on a combination theory planned behaviour (TPB) some elements enhanced stimulus–organism–response (ESOR) theory, which incorporates cognitive, emotional physiological processes organism component. enables development framework facilitates examination purchase environment, moderated attitude (organism). The uniqueness further inclusion analysis platform quality streamer’s (seller’s) influence moderating constructs. These analyses were carried out using data from 901 respondents structured questionnaire, collected over 4-month period. results showed that seller has significant effect mediation attitude, though was affected quality. therefore offers insights Saudi FPC brands, streamers marketing agencies develop optimise sales content strategy.

Язык: Английский

Процитировано

3
Predicting the actual use of artificial intelligence features of Apple Vision Pro using PLS-SEM DOI
Rana Saeed Al-Maroof, Ragad M Tawafak, Waleed Mugahed Al-Rahmi

и другие.

Contemporary Educational Technology, Год журнала: 2025, Номер 17(3), С. ep580 - ep580

Опубликована: Март 27, 2025

Despite the spread of artificial intelligence (AI) tools and applications, Apple Vision Pro (AVP) stands out for its innovative features compared to other types wearable technology. Moreover, traditional glasses have been deficient in incorporating many AI innovations that could enhance user experiences pose new challenges. In response these aspects, this study aims develop a theoretical model by integrating constructs from expectation confirmation (ECM) (expectation satisfaction [SAT]) aspects Uses Gratifications (U&G) theory. The perceived human likeness mediates model. This focuses on educational domain, aiming assess how technology enhances academic environment improves learning outcomes. method used was survey distributed among 134 participants Al Buraimi University College, Oman, two departments: English, linguistics, information consists seven hypotheses emphasize conceptual findings significantly impact predicting actual use (AU) AVP, indicating users’ expectations SAT play pivotal role adoption are closely linked variable likeness. Similarly, factors such as entertainment value, informativeness, lack web irritations influence associated with variable. However, Informativeness gratification failed pass proposal showed negative indicator AU AI. implications drawn results suggest institutions should tailor their courses curricula promote effective

Язык: Английский

Процитировано

0
The Adoption of Technology Acceptance Model in E-commerce with Artificial Intelligence as a Mediator DOI
Aram H. Massoudi, Muslim Najeeb Zaidan

SSRN Electronic Journal, Год журнала: 2025, Номер unknown

Опубликована: Янв. 1, 2025

Язык: Английский

Процитировано

0
ChatGPT and generation ‘Z’: A study on the usage rates of ChatGPT DOI Creative Commons
Md. Asaduzzaman Babu,

Kazi Md. Yusuf,

Lima Nasrin Eni

и другие.

Social Sciences & Humanities Open, Год журнала: 2024, Номер 10, С. 101163 - 101163

Опубликована: Янв. 1, 2024

Язык: Английский

Процитировано

2
Can Multimodal Large Language Models Enhance Performance Benefits Among Higher Education Students? An Investigation Based on the Task–Technology Fit Theory and the Artificial Intelligence Device Use Acceptance Model DOI Open Access
Amany Ahmed Al-Dokhny, Omar A. Alismaiel,

Samia Youssif

и другие.

Sustainability, Год журнала: 2024, Номер 16(23), С. 10780 - 10780

Опубликована: Дек. 9, 2024

The current study highlights the potential of multimodal large language models (MLLMs) to transform higher education by identifying key factors influencing their acceptance and effectiveness. Aligning technology features with educational needs can enhance student engagement learning outcomes. examined role MLLMs in enhancing performance benefits among students, using task–technology fit (T-TF) theory artificial intelligence device use (AIDUA) model. A structured questionnaire was used assess perceptions 550 Saudi university students from various academic disciplines. data were analyzed via structural equation modeling (SEM) SmartPLS 3.0. findings revealed that social influence negatively affected effort expectancy regarding hedonic motivation also related expectancy. for MLLMs. Effort associated T-TF context. In contrast, task characteristics significantly influenced T-TF, which positively impacted both willingness accept strong relationship found between adoption improved benefits. empower educators strategically strategically, driving transformative

Язык: Английский

Процитировано

2
Accelerated Federated Learning Using Self-Adapting Bat Algorithm DOI Creative Commons

Jie Wang,

Chaochao Sun, Peng Yuan

и другие.

Research Square (Research Square), Год журнала: 2024, Номер unknown

Опубликована: Авг. 12, 2024

Abstract Federated learning (FL) is an advanced distributed machine (ML) framework designed to address issues related data silos and privacy. A significant challenge in FL the non-independent identically (Non-IID) nature of client data, resulting slow convergence rate low prediction accuracy for model. To tackle these issues, we propose a scheme based on bat algorithm (FedBat), leveraging echolocation mechanism bats effectively balance global local search capabilities optimizing model weight updates through dynamic adjustments strategy. FedBat also allows adaptive parameter across various datasets. mitigate drift issue, extend by using Jensen-Shannon(JS) divergence quantify difference between models. Clients decide whether upload their models this difference, aiming enhance model's generalization capability minimize communication overhead. Experimental results demonstrate that converges 5 times faster enhances test more than 40% compared FedAvg. The extended mitigates decrease performance reduces costs around 20%. Comparing FedPSO, FedGwo, FedProx shows demonstrates superior terms accuracy. We derive formula expected FedBat, analyze impact parameters performance, establish upper bound evaluate its divergence.

Язык: Английский

Процитировано

0