98047 research outputs found
Sort by
Blended Learning Effectiveness and College Students’ Deep Learning Perceptions: The Community of Inquiry Perspective
No full textEmerging technologies and innovative instructional methods have revolutionized education, making
blended learning the new standard in the artificial intelligence era. However, poor integration of online
and face-to-face learning has led to challenges such as superficial student engagement. This study
developed a Community of Inquiry-based blended learning model and evaluated its effectiveness with
92 college students using a quasi-experimental approach. Over 16 weeks, the experimental group (n =
48) adopted the blended learning model, while the control group (n = 44) used traditional learning
conditions. Learning effectiveness and deep learning perceptions were evaluated, revealing the blended
learning group demonstrated superior learning effectiveness (d = 0.83) and reported higher deep
learning perceptions (η2 = .05–.072) compared to the traditional learning group. These results provide
valuable insights for educators aiming to design blended learning models that foster deep learning and
improve overall learning effectiveness.補正完畢TW
MOSA: Matrix Optimized Self-Attention Hardware Accelerator for Mobile Device
No full textThe Self-Attention mechanism, which lies at the core of Transformer architectures, plays a vital role in capturing long-range dependencies. However, its high computational complexity and significant memory requirements pose major challenges for resource-constrained hardware such as mobile devices. In particular, frequent memory accesses and inefficient matrix multiplication operations often result in performance bottlenecks. Therefore, the development of dedicated hardware accelerators for Self-Attention, focusing on optimizing matrix operations and reducing memory usage, is essential for improving AI processing efficiency on mobile devices. This paper presents a Self-Attention hardware accelerator designed specifically for mobile devices. By optimizing the matrix multiplication process, the accelerator effectively reduces data transmission and memory access frequency, thereby lowering the overall computational complexity. It also reuses intermediate computation results, minimizing frequent memory read and write operations, which significantly reduces memory bandwidth requirements. This data reuse strategy not only cuts down on redundant computations but also greatly enhances computational efficiency. In addition, the accelerator eliminates the need for Key Transpose operations, further simplifying certain computational steps. Compared to traditional algorithms, the idle time is reduced by 66.5%., and signal line transmission costs are reduced by approximately 80%. With an optimized hardware architecture, mobile devices can efficiently support complex deep learning applications, such as speech processing and image recognition.補正完畢國際Osaka, JapanJP
Effects of Participation in Volunteer Activities on CD34+ Hematopoietic Stem Cells
No full textAging populations face health challenges, and volunteering may improve health outcomes. This study investigated
the relationship between volunteering, CD34+ hematopoietic stem cells, and health indicators in older adults.
The sample included 91 participants aged 65 to 75, with 52 reporting prior volunteering involvement. Evaluations
comprised the Geriatric Depression Scale (GDS), Mini-Mental State Examination, Clinical Dementia Rating scale, and
Neuropsychiatric Inventory. Blood analyses measured CD34+ hematopoietic stem cells and CD34+ lymphocytes.
Multiple linear regression assessed variations in CD34+ counts and related metrics among groups, adjusting for
confounding factors. Volunteers exhibited enhanced cognitive functioning and lower depression levels compared to
non-volunteers, as reflected in GDS scores of 4.29 ± 4.18 versus 8.26 ± 5.09 (p < .001). Furthermore, volunteers
had notably elevated CD34+ stem cell and lymphocyte counts, suggesting volunteering positively influences this
health marker. Participation in volunteer activities is linked to improved cognitive abilities, reduced depressive
symptoms, and heightened CD34+ hematopoietic stem cell levels. Findings underscore the potential health benefits
of volunteering for older adults, necessitating further investigation into the mechanisms and enduring effects of
volunteerism on health, with CD34+ cell counts identified as a significant biomarker for cognitive and emotional
health enhancements in this population補正完畢US
代間照顧關係：台灣都會地區成年子女的質性訪談研究
No full text研究目的：「老人的安養與照顧」議題在人口結構逐漸轉變的現代華人社會中越來越受到重視，本研究主要探討華人社會中，成年子女如何因應父母逐漸老去所需面對的照顧需求，並以「性別」、「婚姻」與「代間關係」為研究主要的關注焦點。研究方法：本研究採用具探索性且有助理論建構的質性訪談研究，並透過訪員認識、親友介紹與受訪者介紹等管道招募研究參與者，共邀請34位30∼55歲之間的成年子女接受訪談，研究參與者的性別男女各半，已婚與未婚者各占50%。研究結果：成年子女對年老父母的照顧經驗歸納為三個主題，首先是「代間照顧的方向性」，本研究發現，成年子女似乎是「陪伴」而非「照顧」老年父母，反而是老年父母提供他們各種協助。第二個主題為「家人關係的世代轉變」，研究結果顯示，成年子女照顧父母的經驗中，會出現親子權力反轉的現象，而過去親子關係的良好與否，似乎也左右著成年之後的代間照顧關係。最後是「性別與婚姻的劃界」，未婚女兒似乎較容易被視為理所當然的照顧者，而未婚兒子身上則背負著的家族與經濟期望。此外，女性也會擔心進入婚姻會承擔「媳婦」照顧公婆的角色而失去自我。研究結論：本研究發現代間照顧現象具有強烈的心理意涵與相互性，且照顧意願與方式受性別與婚姻牽引。此外，家庭權力移轉過程中兩代的協商亦考驗著成年子女的能力與智慧。補正完畢TW
工作家庭雙介面之要求、資源與衝突感受之性別差異
No full text本研究目的為探討工作要求、家庭要求、工作資源、與家庭資源等前因變項與職家衝突關連的性別差異。樣本為台灣264名全職工作者，資料分析主要以分群階層迴歸檢驗兩性於職家衝突（包括工作－家庭衝突、家庭－工作衝突）前因之差異。首先，T檢定顯示職家衝突感受均無性別差異。再者，工作負荷與家庭責任是男性與女性職家衝突的顯著預測因子。第三，組織家庭支持文化與主管理念性支持對於男性的雙向職家衝突有顯著預測力，但這兩項工作資源中，僅有「主管理念性支持」對女性工作－家庭衝突有顯著的預測力。第四，「來自配偶的家事協助」是男性工作－家庭衝突的顯著預測因子；然而，對女性來說，「來自父母的家事協助」才可有效降低工作－家庭衝突。是故，組織需了解兩性在職家衝突歷程中的差異，方能協助員工找出最佳因應方式，在職家兩者間取得最終的平衡。補正完畢國內台北市，台灣TW
The use of multicomponent reactions in the development of bis-boronic acids for the detection of β-sialic acid
No full text連結
https://pubs.rsc.org/en/content/articlelanding/2024/ob/d3ob01877fSialic acid (SA) is a naturally occurring monosaccharide found in glycoproteins and glycolipids. Changes in the expression of SA are associated with several diseases; thus, the detection of SA is of great significance for biological research, cancer diagnosis, and treatment. Boronic acid analogs have emerged as a promising tool for detecting sugars such as SA due to its reversible covalent bonding ability. In this study, 11 bis-boronic acid compounds and 2 mono-boronic acid compounds were synthesized via a highly efficient Ugi-4CR strategy. The synthesized compounds were subjected to affinity fluorescence binding experiments to evaluate their binding capability to SA. Compound A1 was shown to have a promising binding constant of 2602 ± 100 M−1 at pH = 6.0. Density Functional Theory (DFT) calculations examining the binding modes between A1 and SA indicated that the position of the boronic acid functional group was strongly correlated with its interaction with SA's α-hydroxy acid unit. The DFT calculations were consistent with the observations from the fluorescence experiments, demonstrating that the number and relative positions of the boronic acid functional groups are critical factors in enhancing the binding affinity to SA. DFT calculations of both S and R configuration of A1 indicated that the effect of the S/R configuration of A1 on its binding with β-sialic acid was insignificant as the Ugi-4CR generated racemic products. A fluorine atom was incorporated into the R2 substituent of A1 as an electron-withdrawing group to produce A5, which possessed a significantly higher capability to bind to SA (Keq = 7015 ± 5 M−1 at pH = 6.0). Finally, A1 and A5 were shown to possess exceptional binding selectivity toward β-sialic acid under pH of 6.0 and 6.5 while preferring to bind with glucose, fructose, and galactose under pH of 7.0 and 7.5.電子
Estimation of value-at-risk for energy commodities via fat-tailed GARCH models
No full textThe choice of an appropriate distribution for return innovations is important in VaR applications owing to its ability to directly affect the estimation quality of the required quantiles. This study investigates the influence of fat-tailed innovation process on the performance of one-day-ahead VaR estimates using three GARCH models (GARCH-N, GARCH-t and GARCH-HT). Daily spot prices of five energy commodities (WTI crude oil, Brent crude oil, heating oil #2, propane and New York Harbor Conventional Gasoline Regular) are used to compare the accuracy and efficiency of the VaR models.
Empirical results suggest that for asset returns that exhibit leptokurtic and fat-tailed features, the VaR estimates generated by the GARCH-HT models have good accuracy at both low and high confidence levels. Additionally, MRSB indicates that the GARCH-HT model is more efficient than alternatives for most cases at high confidence levels. These findings suggest that the heavy-tailed distribution is more suitable for energy commodities, particularly VaR calculation.補正完畢US
The Impact of eWOM on Consumers' Online Purchase Intentions in Vietnam and Taiwan.
No full textThis study explores the impact of electronic word-of-mouth (eWOM) on consumers' online purchase intentions in Vietnam and Taiwan. In the current landscape, online shopping platforms have gained immense popularity as preferred venues for purchasing goods, underscoring the importance of understanding consumers' motivations in opting for online transactions. Data were collected through a questionnaire survey to assess respondents' perceptions of eWOM, including trust in online platforms, platform reviews and ratings, credibility of eWOM, and the recency of eWOM. The findings highlight that the recency of eWOM significantly influences consumers' purchase intentions. These results contribute to the academic fields of e-commerce and consumer behavior, while also providing practical strategic insights for online commerce platforms to better meet the diverse needs and expectations of consumers in varied markets.補正完畢TW
MeTa Learning-Based Optimization of Unsupervised Domain Adaptation Deep Networks
No full textThis paper introduces a novel unsupervised domain adaptation (UDA) method,
MeTa Discriminative Class-Wise MMD (MCWMMD), which combines meta-learning
with a Class-Wise Maximum Mean Discrepancy (MMD) approach to enhance domain adaptation.
Traditional MMD methods align overall distributions but struggle with classwise
alignment, reducing feature distinguishability. MCWMMD incorporates a metamodule
to dynamically learn a deep kernel for MMD, improving alignment accuracy and
model adaptability. This meta-learning technique enhances the model’s ability to generalize
across tasks by ensuring domain-invariant and class-discriminative feature representations.
Despite the complexity of the method, including the need for meta-module
training, it presents a significant advancement in UDA. Future work will explore scalability
in diverse real-world scenarios and further optimize the meta-learning framework.
MCWMMD offers a promising solution to the persistent challenge of domain adaptation,
paving the way for more adaptable and generalizable deep learning models.補正完畢CH