Technische Hochschule Würzburg-Schweinfurt Publikationsserver OPUS
Not a member yet
5300 research outputs found
Sort by
What has become different – participants’ transformations in a Decoding Learning Community
Get PDFDecoding the Disciplines is a way of doing SoTL that focuses on identifying barriers to
student learning and helping students to overcome them. The Decoding Learning
Community is a group of instructors and faculty developers in Germany who intend to
build on that. Within its six years of existence, the group’s work has contributed at least
25 publications to the SoTL-community at large. Viewing this group as an artefact allows
to learn about what people attracts people to Decoding, what helps them to practice
Decoding, and what they gain from doing so. We asked members of this community what
has become different due to their membership. The answers readily fall into patterns. Not
unexpectedly one pattern aligns with the theme of this conference: making connections –
between faculty and developers, between learning theory and teaching practice, between
expertise and student learning. Other patterns, however, have been quite unexpected up
to the point that aspects one might hope members would mention were not covered at all.
(No spoiling information will be given here.) We will connect these findings to related work
on communities of practice. In terms of conference pedagogy, the presentation will
employ elements of immersive theater with members of the learning community dialoging
with the audience from within the auditory
Diffusion Classifier Guidance for Non-robust Classifiers
No full textClassifier guidance is intended to steer a diffusion process such that a given classifier reliably recognizes the generated data point as a certain class. However, most classifier guidance approaches are restricted to robust classifiers, which were specifically trained on the noise of the diffusion forward process. We extend classifier guidance to work with general, non-robust, classifiers that were trained without noise. We analyze the sensitivity of both non-robust and robust classifiers to noise of the diffusion process on the standard CelebA data set, the specialized SportBalls data set and the high-dimensional real-world CelebA-HQ data set. Our findings reveal that non-robust classifiers exhibit significant accuracy degradation under noisy conditions, leading to unstable guidance gradients. To mitigate these issues, we propose a method that utilizes one-step denoised image predictions and implements stabilization techniques inspired by stochastic optimization methods, such as exponential moving averages. Experimental results demonstrate that our approach improves the stability of classifier guidance while maintaining sample diversity and visual quality. This work contributes to advancing conditional sampling techniques in generative models, enabling a broader range of classifiers to be used as guidance classifier
Advances in Self-Organizing Maps, Learning Vector Quantization, Interpretable Machine Learning, and Beyond - Proceedings of the 15th International Workshop, WSOM+ 2024, Mittweida, Germany, July 10-12, 2024
No full text
A critical reflection on modelling approaches for heat pumps and building envelope retrofits in local energy system optimisations
No full textIntegrating heat pumps and adopting building envelope retrofit (BER) measures is vital for decarbonising the built environment. This paper systematically reviews modelling approaches for these technologies within 84 local energy system optimisation studies. Our findings show that most studies simplify heat pump operation using a constant coefficient of performance (COP), perfect modulation, or a fixed heating capacity. While existing methods account for temperature-dependent COPs, only a few studies include more advanced approaches considering additional impacts. Two main approaches for modelling BER measures are identified: pre-defined retrofit sets with exogenous heat demand or an integrated retrofit selection with endogenous heat demand. While the pre-defined approach offers lower computational effort, it limits the BER solution space. The integrated approach achieves the opposite: a wider solution space but higher complexity. Moreover, only two studies consider the impact of BER measures on the heat supply temperature, an essential link between retrofit measures and heat pump efficiency. Based on our qualitative review, we derive recommendations for energy system modellers. For modelling heat pumps, we recommend using a temperature-dependent COP and heating capacity, as well as a lower modulation limit. Regarding BER measures, the integrated approach should be used for smaller scales (single buildings) while the pre-defined sets are suitable for larger scales. Finally, we propose a research agenda to address limitations and gaps in the current modelling approaches, such as addressing the performance gap between existing COP estimations and heat pump field performance and including a more detailed method for the heat supply temperature
Dual Scan head approach for in-situ defect detection in laser powder bed fusion of metals - Dataset
No full textThis dataset contains thermographic data from a study on in-situ defect detection in the Laser Powder Bed Fusion of Metals (PBF-LB/M) process. The data was collected using a novel experimental setup named Synchronized Path Infrared Thermography (SPIT), which employs a dual scan head configuration. One scan head directs the processing laser, while the second scan head positions the measurement field of an infrared (IR) camera. This setup allows for the precise analysis of the cooling behavior of the material decoupled from the immediate laser-material interaction zone.
The experiments were conducted on pre-fabricated stainless steel (EOS StainlessSteel PH1, DIN 14540) samples with embedded, cylindrical subsurface defects of varying diameters. A single layer of metal powder was applied to these samples and then fused by the laser. The dataset includes a series of measurements where process parameters, specifically the volumetric energy density and the laser scanning speed, were systematically varied to assess their influence on defect detection reliability.
The provided data consists of raw thermographic recordings, which capture the surface temperature distribution in the heat-affected zone behind the melt pool. These recordings can be used to identify localized areas of elevated temperature caused by the insulating effect of the subsurface defects, which impede heat transfer into the substrate. This dataset is valuable for researchers working on process monitoring, defect detection algorithms, and the validation of thermal simulations in additive manufacturing
A Computational Approach to Interaction Type Analysis of Music Therapy Improvisations
Get PDFImprovisation in music therapy is a highly complex and diverse form of creativity, offering a wide variety of musical information for music therapists to work with. To address this diversity in research and analysis, it is common to combine a wide range of interdisciplinary scientific approaches. Microanalysis methods in music therapy provide highly insightful results on a detailed musical level in musical improvisation but come at the cost of a time-consuming analysis procedure. The automation of these methods in machine learning environments and the use of the wealth of digitally obtainable musical information in clinical improvisations is highly promising for enabling the efficient use of microanalytic methods in clinical practice. In particular, assessment procedures – the systematic collection and analysis of client information to plan subsequent therapy sessions – can benefit greatly from a microanalytic insight into imitation patterns or entrainment processes as observable in musical instrument digital interface (MIDI) data. However, the automation of microanalytic methods poses a challenge in formalising analytical arguments while at the same time maintaining qualitative validity in a machine learning environment. This article provides an interdisciplinary theoretical framework for the microanalysis of musical data in clinical improvisation that is suitable for computational implementation, leading to the development of an automated analysis tool for further use in research and clinical practice. While a pilot application of the system presented in the article suggests general functionality, future challenges for the training of a supervised classification model have been identified that focus on the need for formalisation of microanalytic arguments and feature development to ensure qualitative validity