There is a huge demand for trying on clothing at home. Recent methods to capture your figure mostly work in a 2d image plane and despite recent improvements of available technology the simulation of clothing is still not satisfactory. Especially the rendering of different clothing sizes is still a major challenge, which is only addressed by COTTON [1]. We propose an improvement to this approach by adding more control over the image generation process. For this we employ a special type of conditional diffusion model, namely ControlNet, and take keypoints of the fashion as conditional input.

Wichtig ist, dass wir KI als Instrument zur Unterstützung begreifen. Der Einsatz von KI aus Marketingsicht und Möglichkeiten für den Bereich Pflege.

Background

It is controversial to what extent afterimages, as distinct visual phenomena, are altered in patients with migraine and whether they have a specific role in migraine pathophysiology.

Objective

The aim of this cross-sectional study was to investigate the duration of afterimages in patients with migraine, migraine with aura (MwA), and migraine without aura (MwoA), compared to healthy controls (HCs).

Methods

Adults with migraine, MwA, and MwoA, diagnosed according to The International Classification of Headache Disorders, third edition criteria and HCs without relevant headache history were included. Initially, factors affecting the experimental setting of testing afterimage latency were determined. Then, afterimage duration was measured in the two study groups (MwA and MwoA) and the HC group. Patient characteristics, intraocular pressure, and relevant comorbid conditions, as well as scales on depressive symptoms (nine-item Patient Health Questionnaire) and headache-specific psychosocial impairment (six-item Headache Impact Test) were recorded. Lastly, the role of different stimulus colors, as well as habituation effects after repeated stimulation, were investigated.

Results

The main study included 174 participants (40 with MwA, 53 with MwoA, and 81 HCs). The duration of the afterimage in patients with MwA was significantly longer than in HCs, at a mean (standard error of the mean [SEM]) of 12.6 (2.6) versus 5.5 ( 0.3) s (p = 0.035), while there was no significant difference between patients with MwoA (mean [SEM] 7.7 [1.6] s; p = 0.510) and HCs. There was also no significant effect of stimulus color on afterimage latency (mean [SEM] red: 8.9 [1.2] s and black: 8.4 [1.2] s).

Conclusion

We found significantly longer afterimage duration in patients with MwA compared to both HCs and patients with MwoA. Furthermore, partially selective stimulation of retinal rods and cones by different stimulus colors had no effect on afterimage duration suggesting a relevant subcortical and/or cortical modulation in migraine aura with increased excitability.

Weinbau dürfte zu den ältesten und traditionsreichsten Branchen der Welt gehören. Politische Umbrüche und – natürlich vor allem in jüngerer Zeit – klimatische und ökologische Veränderungen sorgen, genauso wie technische Innovationen, für stetigen Anpassungsdruck bei der Erzeugung und Vermarktung von Wein. Auch die Digitalisierung der Weinwirtschaft schreitet immer weiter voran. Künstliche Intelligenz (KI) spielt im Weinbau eine zunehmend wichtige Rolle, über alle Glieder der Wertschöpfungskette hinweg.

Große Sprachmodelle wie ChatGPT benötigen große und teure Server und viel Energie. Man kann sie aber quantisieren, sodass sie mit viel weniger Speicher und Strom auskommen und sogar lokal auf einem Smartphone laufen. Wir erklären, warum quantisierte Modelle viel schneller antworten und trotzdem fast so schlau sind wie die großen Originale.

During the corona pandemic, the role of science was seen as something very important in broad sections of society. Unfortunately, this appears to be less the case with regard to climate change. Increasingly, subjective and emotionally based statements are dominating the discussion, which must be viewed extremely critically in view of the complex interactions and extreme effects. On the other hand, great hope is being placed in science to master the complex challenges with the help of artificial intelligence (AI). With the public appearance of generative AI (GenAI) in the form of ChatGPT, a new, partly critical discussion is being held, although AI-based technologies have already made impressive progress as a result of scientific research for many years and numerous developments are already in real use. On the one hand, science is expected to provide insights and recommendations for the responsible use of AI. On the other hand, the use of AI in research work is also discussed critically because it is used, for example, in the application phase of research projects or to evaluate the data obtained and to write down and communicate the results. This gives rise to questions such as “How valid are scientific achievements that are produced with (the help of) AI?” Especially when using generative AI, there is a risk that the own creation process will degenerate due to uncontrolled use, excessive dependency and distortion of results [1], because activities that go beyond routine processes are conveniently left to generative AI, which can be a problem especially when processing complex tasks. An analysis of the authors of around 30 master’s theses from 2023 and 2024 in an international engineering master’s program showed that around 1/3 of the theses are strongly influenced by GenAI results, recognizable by the usual listing form, the text style and the lack of reference to the task. These papers were also among the 1/3 with the lowest grading. Without guidance on the effective use of GenAI, its use does not appear to lead to improved performance, but instead encourages the unthinking copying of text modules.

On the other hand, observations of the use of generative AI in everyday teaching and research show that GenAI can promote complexity competence in particular when used in a targeted manner on the basis of appropriate training. The authors use GenAI in various contexts of their research, increasingly to support the solution of complex problems and tasks in complex environments. The main area of application is water management and increasingly the analysis of urban areas to adapt to water-related challenges caused by climate change. The main focus here is on the massive impact of extreme weather events such as heavy rainfall and prolonged periods of drought on urban and regional infrastructure as well as forest and agricultural areas. These main areas of application are not only highly complex in thematic terms. Solutions must be implemented at urban and municipal level, and here too, the spatial, infrastructural and organizational environment is highly complex. It is against this background that the observations described in the use of generative AI were made, the theoretical framework described in this article was created and the hypothetical analysis was carried out, for which empirical evidence is still pending, but in preparation.

Despite being a specific sustainable development goal (SDG), the role of consumers for sustainable consumption is still ambiguous. This is exemplified by a vast amount of research on the attitude-intention-behavior gap, which generally describes consumers’ failures to behave as sustainable as theoretically predicted. Recent reviews have prompted further investigations beyond the existing literature on factors influencing this gap. We contribute to this call by quantitatively investigating five antagonistic dimensions – both intrapsychic and situation-related – of smartphone usage and sustainable consumer behavior in Germany (n=800). Our results indicate two novel concepts. Emotional connection – i.e., consumers’ connections with the consumption experience – can either promote or prevent sustainable behavior, while exploration-driven consumerism – i.e., new purchases due to exploration tendencies – typically attenuates sustainable behavior. This illustrates how and when sustainability is outweighed by other consumer attitudes. We contextualize these results and conclude our study by highlighting limitations and further research opportunities.

Seien Sie dabei, wenn wir am Dienstag, den 17. Dezember, einen aufschlussreichen Webcast zum Thema „CX und KI - eine deutsche Perspektive“ anbieten! 🌟 Entdecken Sie die einzigartige Sichtweise auf Customer Experience (CX) und Künstliche Intelligenz (KI) aus Deutschland. Unsere Experten werden sich mit folgenden Themen befassen: Die deutsche CX-Perspektive und ihre Unterschiede zu Europa und den USA. Der deutsche KI-Ansatz mit Fokus auf Branchen, Datenoptimierung und Qualitätsmanagement. KI-Anwendungen aus der Praxis in Vertrieb, Marketing und dem öffentlichen Sektor. Die Landschaft der KI-Startups in Deutschland und der DACH-Region. Herausforderungen und Chancen der KI-Finanzierung, Regulierung und Nachhaltigkeit. Verpassen Sie nicht die Gelegenheit, wertvolle Einblicke zu gewinnen und sich mit Referenten und Experten der Branche auszutauschen! Join us for an insightful webcast on "CX and AI - A German Perspective" on Tuesday, December 17th! 🌟 Explore the unique viewpoints on Customer Experience (CX) and Artificial Intelligence (AI) from Germany. Our experts will delve into: The German CX perspective and its differences from Europe and the USA. The German AI approach, focusing on sectors, data optimization, and quality management. Real-world AI applications in sales, marketing, and the public sector. The landscape of AI startups in Germany and the DACH region. Challenges and opportunities in AI financing, regulation, and sustainability. Don't miss this opportunity to gain valuable insights and engage with industry speakers and experts!

Literature reviews play a crucial role in Information Systems (IS) research. However, scholars have expressed concerns regarding the reproducibility of their results and the quality of documentation. The involvement of human reproducers in these reviews is often hindered by the time-consuming nature of the procedures. The emergence of Large Language Models (LLMs) seems promising to support researchers and to enhance reproducibility. To explore this potential, we conducted experiments using various LLMs, focusing on abstract scanning, and have presented initial evidence suggesting that the application of LLMs in structured literature reviews could assist researchers in refining and formulating rules for abstract scanning. Based on our preliminary findings, we identify potential future research directions in this research in progress paper.

Angesichts des enormen Veränderungspotenzials, das ChatGPT, Perplexity & Co für die Arbeitswelt bergen, stellt sich die Frage, wie man diese Instrumente der generativen KI sinnvoll in den Arbeitsalltag integriert. Einen entsprechenden Analyseansatz bietet das Kategorisierungsmodell der „Cyborgs und Zentauren“.

As the output quality of neural networks in the fields of automatic speech recognition (ASR) and text-to-speech (TTS) continues to improve, new opportunities are becoming available to train models in a weakly supervised fashion, thus minimizing the manual effort required to annotate new audio data for supervised training. While weak supervision has recently shown very promising results in the domain of ASR, speech synthesis has not yet been thoroughly investigated regarding this technique despite requiring the equivalent training dataset structure of aligned audio-transcript pairs.
In this work, we compare the performance of TTS models trained using a well-curated and manually labeled training dataset to others trained on the same audio data with text labels generated using both grapheme- and phoneme-based ASR models. Phoneme-based approaches seem especially promising, since even for wrongly predicted phonemes, the resulting word is more likely to sound similar to the originally spoken word than for grapheme-based predictions.
For evaluation and ranking, we generate synthesized audio outputs from all previously trained models using input texts sourced from a selection of speech recognition datasets covering a wide range of application domains. These synthesized outputs are subsequently fed into multiple state-of-the-art ASR models with their output text predictions being compared to the initial TTS model input texts. This comparison enables an objective assessment of the intelligibility of the audio outputs from all TTS models, by utilizing metrics like word error rate and character error rate.
Our results not only show that models trained on data generated with weak supervision achieve comparable quality to models trained on manually labeled datasets, but can outperform the latter, even for small, well-curated speech datasets. These findings suggest that the future creation of labeled datasets for supervised training of TTS models may not require any manual annotation but can be fully automated.

There are many publications talking about the biases to be found in in generative AI solutions like large language models (LLMs, e.g., Mistral) or text-to-image models (T2IMs, e.g., Stable Diffusion). However, there is merely any publication to be found that questions what kind of behavior is actually desired, not only by a couple of researchers, but by society in general. Most researchers in this area seem to think that there would be a common agreement, but political debate in other areas shows that this is seldom the case, even for a single country. Climate change, for example, is an empirically well-proven scientific fact, 197 countries (including Germany) have declared to do their best to limit global warming to a maximum of 1.5°C in the Paris Agreement, but still renowned German scientists are calling LLMs biased if they state that there is human-made climate change and humanity is doing not enough to stop it. This trend is especially visible in Western individualistic societies that favor personal well-being over common good. In this article, we are exploring different aspects of biases found in LLMs and T2IMs, highlight potential divergence in the perception of ethically desirable outputs and discuss potential solutions with their advantages and drawbacks from the perspective of society. The analysis is carried out in an interdisciplinary manner with the authors coming from as diverse backgrounds as business information systems, political sciences, and law. Our contribution brings new insights to this debate and sheds light on an important aspect of the discussion that is largely ignored up to now.

Vergnügungsparks stehen vor der Herausforderung, sich in einer zunehmend digitalisierten Welt weiterzuentwickeln, um ihre Attraktivität für ein breites Publikum zu sichern. Digitale Technologien, insbesondere Virtual Reality (VR), eröffnen hier neue Perspektiven. Sie verändern nicht nur das Besuchererlebnis, sondern haben auch einen signifikanten Einfluss auf den Geschäftserfolg.

In this paper we present a stable-diffusion based zero-shot approach to realistically transform the image of a

human body into a more fit version of that depicted person. Therefore we combine a modified stable diffusion

model with inpainting techniques and incorporated constraints. We introduce a prototype which allows users to

upload a photo and visualize a more fit version of themselves. We evaluated our approach in various experiments

and focused on the applicability and effectiveness of these techniques, with attention to gender-specific results.

This work contributes to the fields of computer vision and generative AI by demonstrating practical applications

and identifying areas for improvement in realistic body transformation visualizations.