Refine
Document Type
- Article (reviewed) (3) (remove)
Language
- English (3)
Has Fulltext
- no (3)
Is part of the Bibliography
- yes (3) (remove)
Keywords
- Deep Leaning (1)
- Just-in-time Learning (1)
- Manufacturing Processes (1)
- Media Streaming Services (1)
- Mobile Devices (1)
- Pull Service (1)
- Push Service (1)
- SMIL (1)
- Virtual experience (1)
- Virual Reality (1)
Institute
Open Access
- Bronze (3) (remove)
Virtual reality (VR) offers the opportunity to create virtual worlds that could replace real experiences. This research investigates the influence of user motivation, temporal distance and experience type on the satisfaction with the VR experience, and the degree of acceptance of a VR experience as a substitute for a real experience. The results suggest that the degree of acceptance of a VR experience as a substitute for a real experience is higher for passive VR experiences compared to active VR experiences. Furthermore, the results support the assumption that users are more satisfied with passive VR experiences.
Neural networks tend to overfit the training distribution and perform poorly on out-ofdistribution data. A conceptually simple solution lies in adversarial training, which introduces worst-case perturbations into the training data and thus improves model generalization to some extent. However, it is only one ingredient towards generally more robust models and requires knowledge about the potential attacks or inference time data corruptions during model training. This paper focuses on the native robustness of models that can learn robust behavior directly from conventional training data without out-of-distribution examples. To this end, we study the frequencies in learned convolution filters. Clean-trained models often prioritize high-frequency information, whereas adversarial training enforces models to shift the focus to low-frequency details during training. By mimicking this behavior through frequency regularization in learned convolution weights, we achieve improved native robustness to adversarial attacks, common corruptions, and other out-of-distribution tests. Additionally, this method leads to more favorable shifts in decision-making towards low-frequency information, such as shapes, which inherently aligns more closely with human vision.
This paper presents a streaming-based E-Learning environment where closer integration between learning and work is achieved by integrating multimedia services into manufacturing processes. It contains a comprehensive and detailed explanation of the proposed E-Learning streaming framework, especially the adaption of streaming services to mobile environments. We first analyze several scenarios where E-Learning streaming services can be integrated into manufacturing processes. To allow systematic and tailor-made integration, we develop a model and a specification language for E-Learning streaming services and apply the model using practical scenarios from real manufacturing processes. Adaption of multimedia streaming services to mobile devices is discussed based on Synchronized Multimedia Integration Language (SMIL). Last, we comment on the benefits of using E-Learning streaming services as part of manufacturing processes and analyze the acceptance of the developed system. The key components of our E-Learning environment are 1) an xml based streaming service specification language, 2) adaption of multimedia E-Learning services to mobile environments, and 3) Web Services for searching, registration, and creation of E-Learning streaming services.