Refine
Year of publication
- 2024 (3) (remove)
Institute
Has Fulltext
- no (3)
Language
- English (3)
Document Type
- Preprint (3) (remove)
Keywords
- Dynamic modeling (1)
- ESATAN-TMS (1)
- Lunar Surface (1)
- Thermal Model (1)
- Thermal analysis (1)
Zugriffsart
- weltweit (3)
Is part of the Bibliography
- no (3)
This paper presents a thermal simulation environment for moving objects on the lunar surface. The goal of the thermal simulation environment is to enable the reliable prediction of the temperature development of a given object on the lunar surface by providing the respective heat fluxes for a mission on a given travel path. The user can import any object geometry and freely define the path that the object should travel. Using the path of the object, the relevant lunar surface geometry is imported from a digital elevation model. The relevant parts of the lunar surface are determined based on distance to the defined path. A thermal model of these surface sections is generated, consisting of a porous layer on top and a denser layer below. The object is moved across the lunar surface, and its inclination is adapted depending on the slope of the terrain below it. Finally, a transient thermal analysis of the object and its environment is performed at several positions on its path and the results are visualized. The paper introduces details on the thermal modeling of the lunar surface, as well as its verification. Furthermore, the structure of the created software is presented. The robustness of the environment is verified with the help of sensitivity studies and possible improvements are presented.
Analyzing electroencephalographic (EEG) time series can be challenging, especially with deep neural networks, due to the large variability among human subjects and often small datasets. To address these challenges, various strategies, such as self-supervised learning, have been suggested, but they typically rely on extensive empirical datasets. Inspired by recent advances in computer vision, we propose a pretraining task termed "frequency pretraining" to pretrain a neural network for sleep staging by predicting the frequency content of randomly generated synthetic time series. Our experiments demonstrate that our method surpasses fully supervised learning in scenarios with limited data and few subjects, and matches its performance in regimes with many subjects. Furthermore, our results underline the relevance of frequency information for sleep stage scoring, while also demonstrating that deep neural networks utilize information beyond frequencies to enhance sleep staging performance, which is consistent with previous research. We anticipate that our approach will be advantageous across a broad spectrum of applications where EEG data is limited or derived from a small number of subjects, including the domain of brain-computer interfaces.
Preprint: Detecting sexism in German online newspaper comments with open-source text embeddings
(2024)
Sexism in online media comments is a pervasive challenge that often manifests subtly, complicating moderation efforts as interpretations of what constitutes sexism can vary among individuals. We study monolingual and multilingual open-source text embeddings to reliably detect sexism and misogyny in Germanlanguage online comments from an Austrian newspaper. We observed classifiers trained on text embeddings to mimic closely the individual judgements of human annotators. Our method showed robust performance in the GermEval 2024 GerMS-Detect Subtask 1 challenge, achieving an average macro F1 score of 0.597 (4th place, as reported on Codabench). It also accurately predicted the distribution of human annotations in GerMS-Detect Subtask 2, with an average Jensen-Shannon distance of 0.301 (2nd place). The computational efficiency of our approach suggests potential for scalable applications across various languages and linguistic contexts.