Probability output of multi-class support vector machines

A novel approach to interpret the outputs of multi-class support vector machines is proposed in this paper. Using the geometrical interpretation of the classifying heperplane and the distance of the pattern from the hyperplane, one can calculate the posterior probability in binary classification case. This paper focuses on the probability output in multi-class phase where both the one-against-one and one-against-rest strategies are considered. Experiment on the speaker verification showed that this method has high performance.     

Understanding the influence of teachers’ cognitive and affective responses upon school inspection feedback acceptance

Despite the developmental perspective of school inspections, teachers in inspected schools are not always willing to accept the school inspection’s feedback for their further improvement of teaching and learning processes. Literature distinguishes several aspects of feedback that stimulate or hinder the acceptance of feedback, such as recipient’s cognitive and affective responses to feedback. This study investigates teachers’ cognitive and affective responses to school inspection feedback in relation to feedback acceptance. It draws on data from 21 in-depth interviews with teachers in eight primary schools. We found that positive perceptions of the inspectors’ credibility enhance teachers’ feedback acceptance. This is also the case for positive, clear feedback. Under these circumstances, emotions of joy, happiness and relief are expressed. Conversely, respondents tend to reject feedback when inspectors are perceived to be inadequately informed, arrogant or disrespectful. When negative feedback is rated as unfair, negative emotions, such as anger and sadness, interfere with feedback acceptance. In essence, we conclude that both feedback content and feedback source characteristics are decisive in the acceptance of process. From a practical perspective, the findings suggest there is a need to build on supportive relationships between teachers and school inspectors.     

Initial Evidence on the Influence of College Student Engagement on Early Career Earnings

Using data from three waves of longitudinal surveys of the 2001 freshman cohort of the Gates Millennium Scholars (GMS) program, this study examined linkages between student engagement in college activities and early career earnings in the labor market. The results indicated that social engagement was positively related to early career earnings of college graduates while academic engagement was not. When considering students who majored in STEM fields versus those in non-STEM fields, results showed academic engagement to be positively related to early career earnings for non-STEM students but slightly negatively related to earnings in the labor market for STEM students. Social engagement was positive for graduates in the STEM fields but not significant for non-STEM students. Altogether, results show a complex relationship between student engagement and early career earnings. Implications for research and institutional policy are discussed.     

Reflection on the education policy orientation in post-May 31 reform in Korea

For the last 70 years, since the establishment of the Republic of Korea, Korean education has achieved universal expansion of educational opportunity from elementary to secondary to higher education. Planning, centralized policy making, top-down implementation, and administrative control had been the standards of the first few decades of Korean education. The so-called May 31 education reform implemented in 1995 made a turning point for these approaches to education policy and administration by highlighting autonomy, openness, diversity, accountability, consumer centeredness, market control, and governance for quality education. However, these government-driven education reform initiatives have resulted in limiting the expected outcomes. In this paper, the authors call for a new approach to education policy and administration and propose supporting school innovation, empowering an internal accountability system in governance. The authors also suggest a search for normative values for democratic communitarianism as a new philosophical foundation of education policy framework, beyond instrumental values of education.     

The influence of Work-Integrated Learning and paid work during studies on graduate employment and underemployment

To enhance employability and improve the career prospects of graduating students, this study explores the influence of practical experience on graduate employment outcomes in an Australian setting. To develop our understanding of the relative benefit of different forms of practical experience, the study evaluates the influence of both Work-Integrated Learning (WIL) and paid work in the final year of study on graduate employment and underemployment. Two samples are used, N?=?628 and N?=?237, to evaluate institutional data on practical experience combined with national data on graduate employment outcomes. Findings indicate that participating in WIL does not produce an increase in full-time employment rates. There is some evidence to suggest that it could lead to higher quality, relevant employment in both the short and long term. Paid employment during the final year of undergraduate study produced higher full-time employment rates, but had little effect on underemployment. Findings will help to inform stakeholders of the relative benefit of curricular and extra-curricular work experience and contribute to the dearth of empirical evidence on the value of activities designed to improve graduate employment prospects. This is particularly important given growth in the supply of graduates, concerns for credentialism, soft graduate labour markets and global economic weakening.     

Methylenetetrahydrofolate Reductase C677T Polymorphism and Recurrent Pregnancy Loss Risk in Asian Population: A Meta-analysis

The C677T polymorphism of the methylenetetrahydrofolate reductase (MTHFR) gene was implicated to be associated with thrombophilia due to its role in catalyzing the formation of 5-methylenetetrahydrofolate, a co-substrate for the conversion of homocysteine to methionine. Several case–control studies were investigated MTHFR C677T polymorphism as risk for recurrent pregnancy loss (RPL). These studies rendered contradictory results, some indicating that the polymorphism is associated with the risk of RPL whereas others concluded there is no association. To shed light on these inconclusive findings, a meta-analysis of all available studies published from Asian population relating the C677T polymorphism to the risk of RPL was conducted. The following electronic databases were searched without language restrictions: PubMed, Google Scholars, Elsevier and Springer Link up to December, 2015. Meta-analysis was performed using MetaAnalyst and Mix version 1.7. Meta-analysis results suggested that MTHFR C677T polymorphism contributed to the increased RPL risk in Asian population using all five genetic models (for T vs. C: OR 1.35, 95 % CI 1.09–1.68, p = 0.009; for TT + CT vs. CC: OR 1.44, 95 % CI 1.14–1.82, p = 0.006; for CT vs. CC: OR 1.39, 95 % CI 1.07–1.8, p = 0.01; for TT vs. CC: OR 1.79, 95 % CI 1.23.2.6, p = 0.007; for TT vs. CT + CC: OR 1.61, 95 % CI 1.02–2.56, p = 0.04). In conclusion, this meta-analysis demonstrates a strong association between the MTHFR C677T variant and RPL in Asian population and raising the importance of the use of folate in its treatment and prevention.     

Elementary Teachers’ Selection and Use of Visual Models

As science grows in complexity, science teachers face an increasing challenge of helping students interpret models that represent complex science systems. Little is known about how teachers select and use models when planning lessons. This mixed methods study investigated the pedagogical approaches and visual models used by elementary in-service and preservice teachers in the development of a science lesson about a complex system (e.g., water cycle). Sixty-seven elementary in-service and 69 elementary preservice teachers completed a card sort task designed to document the types of visual models (e.g., images) that teachers choose when planning science instruction. Quantitative and qualitative analyses were conducted to analyze the card sort task. Semistructured interviews were conducted with a subsample of teachers to elicit the rationale for image selection. Results from this study showed that both experienced in-service teachers and novice preservice teachers tended to select similar models and use similar rationales for images to be used in lessons. Teachers tended to select models that were aesthetically pleasing and simple in design and illustrated specific elements of the water cycle. The results also showed that teachers were not likely to select images that represented the less obvious dimensions of the water cycle. Furthermore, teachers selected visual models more as a pedagogical tool to illustrate specific elements of the water cycle and less often as a tool to promote student learning related to complex systems.     

Students’ Views of Scientific Models and Modeling: Do Representational Characteristics of Models and Students’ Educational Levels Matter?

The aim of this study was to examine the potential impact of the representational characteristics of models and students’ educational levels on students’ views of scientific models and modeling (VSMM). An online multimedia questionnaire was designed to address three major aspects of VSMM, namely the nature of models, the nature of modeling, and the purpose of models. The three scales of representational characteristics included modality, dimensionality, and dynamics. A total of 102 eighth graders and 87 eleventh graders were surveyed. Both quantitative data and written responses were analyzed. The influence of the representational characteristics seemed to be more salient on the nature of models and the purpose of models. Some interactions between the educational levels and the representational characteristics showed that the high school students were more likely to recognize textual representations and pictorial representations as models, while also being more likely to appreciate the differences between 2D and 3D models. However, some other differences between educational levels did not necessarily suggest that the high school students attained more sophisticated VSMM. For instance, in considering what information should be included in a model, students’ attention to particular affordances of the representation can lead to a more naive view of modeling. Implications for developing future questionnaires and for teaching modeling are suggested in this study.