240106 research outputs found
Sort by
Generative AI in language learning – text and image generation in L2 an L3
No full textGenerative AI in language learning – text and image generation in L2 an L3Bente Meyer, Aalborg University, Copenhagen, and Annette Søndergaard Gregersen, University College, CopenhagenThe introduction of generative AIs to education has reinitiated discussions of agency, centered e.g. on how learners collaborate with technology to produce text and written language and how this affects practices of teaching and learning (Godwin-Jones 2024, Thorne 2024). In this paper we investigate how generative AIs become part of language teaching through a project in which teacher students in English, German and French were introduced to generative AIs and experimented with ways in which they could be used in classrooms as part of their pre-service teaching in schools. Our research draws on workshops and group interviews with teachers and observation data in classrooms (4th and 6th formers) where teacher students taught as part of their pre-service training. In the project both independent platforms such as ChatGPT and Copilot and integrated AI functions in Padlet (image generation through “I can’t draw”) were used. We also chose to work with SkoleGPT, a curated, data-safe GPT developed for use in schools by the participating teacher college (skolegpt.dk).The paper draws on sociomaterial perspectives to understand how classroom didactics are formed by GenAIs and how this affects pupils’ agency in terms of how they produce text and images in language learning (Toohey 2018, Meyer & Gregersen, forthcoming). Our analysis centers on a specific example of how French was taught to 6th formers using text and image generation in a complex material set-up that connected natural artefacts with AI technologies to teach children French vocabulary.Our preliminary findings suggest that Generative AIs become part of language teaching and learning in different ways, depending for instance on the status of writing in respectively L2s and L3s. As our research has involved both image and text generation with AI, our data contribute with specific knowledge of how these can be used in teaching and how they affect agencies and ecologies of language learning References: Godwin-Jones, R. (2024) Distributed agency in second language learning and teaching through generative AI. Language Learning & Technology. Volume 28, 2. p..5-31 https://doi.org/10125/73570 Meyer, B. & Gregersen, A. (forthcoming) Collaborative language learning through generative AI: the case of French. Academic Quarter, Special issue: Generative AI in Cllaborative Learning Environments. vol 31Thorne, S. L. (2024). Generative artificial intelligence, co-evolution, and language education. Modern Language Journal, vol 108, 2, p. 567-572 https://doi-org.zorac.aub.aau.dk/10.1111/modl.12932 Toohey, K. (2018) Learning English at School. Identity, Socio-material Relations and Classroom Practice. Multilingual Matters<br/
Stability Analysis of Wireless Power Transfer System with Resonant Current RMS Inner Loop Under Communication Failure
No full text
Power Cycling Testing for Power Semiconductor Switches:Methods, Standards, Limitations, and Outlooks
No full textReliability is a critical performance metric for power semiconductor switches and power electronic systems. Yet guidance on how to test and quantify that reliability is fragmented in the existing literature, particularly with the rapid adoption of wide-bandgap (WBG) devices and novel packaging technologies. This review brings guidance on what designers, reliability engineers, and researchers need to know about power cycling testing (PCT). We provide three major contents: first, introducing how new materials and packaging shift dominant failure mechanisms; second, comparing the main PCT standards joint electron device engineering council (JEDEC), automotive electronics council (AEC), international electrotechnical commission (IEC), and automotive qualification guideline (AQG) and explaining why the “test-to-fail” standard principle is overtaking legacy “test-to-pass” rules; and third, summarizing the unique challenges and existing solutions of applying PCT methods to WBG and ultra-WBG devices. Notably, to the best of the authors’ knowledge, this is the first in-depth analysis of the newly released IEC 60749-34:2025 and AQG 324:2025, benchmarked against their earlier editions. Moreover, by collecting more than 200 testing samples from the existing literature, we also offer the first generic lifetime model that spans Si, SiC, multiple bond-wire materials, and die-attach technologies. Finally, the limitations and associated open questions are discussed to identify future research opportunities.</p
Differentiable Predictive Control for Power Electronic Systems
No full textThis letter presents the first implementation of differentiable predictive control (DPC) in power electronic systems. By embedding the predictive control cost function directly into a differentiable neural network policy, the proposed approach eliminates the need for labeled control data and online optimization solvers. The control policy is trained solely from system trajectories and deployed on a shallow network suitable for an embedded execution. Experimental validation demonstrates that DPC achieves an accurate reference tracking, strong generalization under unseen conditions, and significantly lower computational cost compared to conventional model predictive control, particularly for extended prediction horizons. These results highlight the potential of DPC as a data-efficient and computationally scalable alternative for advanced power converter control.</p
The Impact of Graph Structure, Cluster Centroid and Text Review Embeddings on Recommendation Methods
Full text linkIt is generally accepted that collaborative information is important for the performance of recommender systems. It is also generally accepted that if this information is sparser, it impacts recommendation systems negatively. Various approaches have tried to lift this problem by employing side information. However, global patterns that can be provided by clusters of similar items and users or even additional information such as text are often not used together with collaborative information. We study the impact of integrating clustering embeddings, review embeddings, and their combinations with embeddings obtained by a recommender system. We study the performance of this approach across various state-of-the-art recommender system algorithms including graph-based methods. We highlight that graph structures are important with sparser datasets and both, in knowledge graphs with side information as well as in collaborative bipartite graphs. In less sparse datasets, a collaborative bipartite graph is usually sufficient. We also highlight that the improvement of recommendation performance through clustering, particularly evident when combined with review embeddings is most visible on sparser data, while on less sparse data incorporating review embeddings may be sufficient when combined with one of the graph-based methods, or otherwise when combined with clustering in other methods
What is it like to be an AI?
No full text"In a polemical and speculative essay, we update Thomas Nagel’s classic question (What is it like to be a bat?) for our Artificial Intelligence [AI] age. We explore the phenomenological ontology (the nature of experiential being) of AI, speculating on what a conscious AI’s conception of self and non-self might be while questioning current assumptions and marketing hype. Everything that is AI is built on an anthropocentric foundation; the foundation of its ontology derives from generalised human knowledge and our own phenomenology and is filtered further through human-engineered algorithms. With this in mind, we must assume that any potential phenomenological ontology of an AI would be restricted to that of humans. Furthermore, following Nagel's argument that we cannot possibly know what it is like to be a bat, restricted as we are by an anthropocentric embodiment and phenomenology, we must also conclude that an AI, likewise, cannot possibly know what it is like to be a bat, or indeed any other species: all claims to the contrary must be viewed as wishful anthropomorphism. In order to present our arguments, our essay will cover a range of topics all focussed on the question What is it like to be an AI? These include: consciousness; embodiment; presence; creativity; anthropomorphism and anthropodenial; bias; and artificial otherness. We do not deny the possibility of an AI one day having a conception of self with a phenomenological ontology that is different to that of a human being. Rather, our purpose is to point out that, should this occur, we ourselves will be incapable of knowing, at least fully, what it is like to be an AI. But would an AI, with an anthropocentric ontology, even an AIcentric ontology, be capable of knowing what it is like to be a human? As AI increasingly pervades and directs our lives, what might be the ethical implications of allowing an artificial other, with a potentially unknowable phenomenological ontology, this level of power?
Improved wound healing by dual inhibition of miR-146a-5p and miR-29a-3p supports a network action of dysregulated miRNAs in diabetic skin
No full textAims/hypothesis: Upregulation of miR-146a-5p and miR-29-3p is observed in chronic non-healing wounds in diabetes. Their single or combined inhibition's molecular and cellular effects were assessed in human keratinocytes (HaCaT cells) and in vivo using a mouse model of type 1 diabetes. Methods: As primary outcomes, we screened for proteome changes in HaCaT cells by LC-MS/MS after transfection with miR-146a-5p or miR-29a-3p inhibitors individually or in combination and following stimulation with TNF-α. Moreover, as a secondary outcome, we collected the data and cryopreserved and paraffin-embedded skin biopsies to estimate the tissue response to miRNA inhibition using immunofluorescence and histological analysis. Cryopreserved biopsies were also used for the LC-MS/MS proteome profiling to identify targets and cellular pathways involved in observed tissue changes. Results: We identified a panel of extracellular matrix proteins, mainly laminins, whose levels changed after transfection with miR-146a-5p or miR-29a-3p inhibitors in HaCaT cells, counteracting TNF-α effects. There was a difference in wound closure rate in vivo between the dual inhibition of miR-146a-5p and miR-29a-3p and scramble controls on day 8 (p<0.01) and day 9 (p<0.05), although not at day 10. Histological analysis at day 10 shows a loose papillary layer in the scramble inhibition group, indicating incomplete wound closure compared with dual miRNA inhibition. Moreover, the dual action of the inhibitors decreased inflammation at day 3 and day 10 (both p<0.001) and reactive oxygen species formation (p<0.01) 3 days post wounding, while increasing the angiogenesis on day 3 (p<0.01) and day 10 (p<0.001). This was consistent with cytoskeletal rearrangements and collagen alterations observed in proteome profiling. Conclusions/interpretation: These findings demonstrate that dual inhibition of miR-146a-5p and miR-29a-3p in vitro synergises in a bidirectional manner, resulting either in intermediate effects or in cancelling each other’s activity for the levels of specific proteins of basal lamina that impair proliferation and cell motility, compared with the individual inhibitors. Topical supplementation of miR-146a-5p and miR-29a-3p inhibitors to diabetic mouse wounds resulted in a reduction in wound size on days 8 and 9, which correspond to the later stages of healing, but did not lead to complete healing by day 10. However, dual inhibition demonstrates favourable effects on high oxidative stress, elevated inflammation and poor angiogenesis. These effects are superior to single miRNA inhibition, suggesting that combined miRNA inhibition could be a promising therapeutic strategy for diabetic wound healing. Nevertheless, further studies in humans are warranted.</p
Coaching With and For Creativity
No full textThe dynamic landscape of contemporary sport demands that coaches move beyondreplicating established “best practices” to cultivate “next practices” that integrate peak performance with holistic athlete well-being. This chapter introduces the Coaching With and For Creativity framework, a socio-cultural approach designed to guide coach development in enhancing both coaches’ and athletes’ creative potential. Grounded in Glăveanu’s 5A model of creativity, the framework positions creativity as a process of exploration and expansion of possibilities, emphasising the reciprocal and relational dynamics between coach and athlete. We outline three central pathways: expanding creativity-related skills (e.g., openness, perspective-taking, tolerance for ambiguity), enriching environments through creativity-supportive instructions and movement prompts and fostering exploration as a co-creation strategy. Drawing on research and applied case studies, including the Australian National Generation 2032 Coach Program, we demonstrate how embodied creative activities can strengthen coaches’ willingness to take risks, loosen control, and design environments thatinvite athletes to explore, innovate, and grow. Ultimately, Coaching With and For Creativity positions coaches as designers of enriched, dynamic learning systems where creativity is not an individual trait but a distributed, relational process. This framework offers a pragmatic contribution to advancing coach development and nurturing transformative sport environments