<?xml version="1.0" encoding="utf-8"?>
<export-example>
  <doc>
    <id>8393</id>
    <completedYear/>
    <publishedYear>2025</publishedYear>
    <thesisYearAccepted/>
    <language>eng</language>
    <pageFirst/>
    <pageLast/>
    <pageNumber>11</pageNumber>
    <edition/>
    <issue/>
    <volume/>
    <type>preprint</type>
    <publisherName/>
    <publisherPlace/>
    <creatingCorporation/>
    <contributingCorporation/>
    <belongsToBibliography>0</belongsToBibliography>
    <completedDate>--</completedDate>
    <publishedDate>--</publishedDate>
    <thesisDateAccepted>--</thesisDateAccepted>
    <title language="eng">Tackling fake images in cybersecurity - interpretation of a StyleGAN and lifting its black-box</title>
    <abstract language="eng">In today's digital age, concerns about the dangers of AI-generated images are increasingly common. One powerful tool in this domain is StyleGAN (style-based generative adversarial networks), a generative adversarial network capable of producing highly realistic synthetic faces. To gain a deeper understanding of how such a model operates, this work focuses on analyzing the inner workings of StyleGAN's generator component. Key architectural elements and techniques, such as the Equalized Learning Rate, are explored in detail to shed light on the model's behavior. A StyleGAN model is trained using the PyTorch framework, enabling direct inspection of its learned weights. Through pruning, it is revealed that a significant number of these weights can be removed without drastically affecting the output, leading to reduced computational requirements. Moreover, the role of the latent vector -- which heavily influences the appearance of the generated faces -- is closely examined. Global alterations to this vector primarily affect aspects like color tones, while targeted changes to individual dimensions allow for precise manipulation of specific facial features. This ability to finetune visual traits is not only of academic interest but also highlights a serious ethical concern: the potential misuse of such technology. Malicious actors could exploit this capability to fabricate convincing fake identities, posing significant risks in the context of digital deception and cybercrime.</abstract>
    <identifier type="doi">10.48550/arXiv.2507.13722</identifier>
    <identifier type="arxiv">arXiv:2507.13722</identifier>
    <identifier type="urn">urn:nbn:de:bvb:898-opus4-83930</identifier>
    <enrichment key="opus.source">publish</enrichment>
    <licence>Creative Commons - CC BY-NC-SA - Namensnennung - Nicht kommerziell -  Weitergabe unter gleichen Bedingungen 4.0 International</licence>
    <author>Julia Laubmann</author>
    <author>Johannes Reschke</author>
    <subject>
      <language>eng</language>
      <type>uncontrolled</type>
      <value>Generative adversarial networks</value>
    </subject>
    <subject>
      <language>eng</language>
      <type>uncontrolled</type>
      <value>GAN</value>
    </subject>
    <subject>
      <language>eng</language>
      <type>uncontrolled</type>
      <value>StyleGAN</value>
    </subject>
    <subject>
      <language>eng</language>
      <type>uncontrolled</type>
      <value>Artificial Intelligence</value>
    </subject>
    <subject>
      <language>eng</language>
      <type>uncontrolled</type>
      <value>Deep Learning</value>
    </subject>
    <collection role="institutes" number="FakEI">Fakultät Elektro- und Informationstechnik</collection>
    <collection role="othforschungsschwerpunkt" number="16315">Information und Kommunikation</collection>
    <thesisPublisher>Ostbayerische Technische Hochschule Regensburg</thesisPublisher>
    <file>https://opus4.kobv.de/opus4-oth-regensburg/files/8393/conference_paper.pdf</file>
  </doc>
  <doc>
    <id>7133</id>
    <completedYear/>
    <publishedYear>2024</publishedYear>
    <thesisYearAccepted/>
    <language>eng</language>
    <pageFirst>2871</pageFirst>
    <pageLast>2880</pageLast>
    <pageNumber/>
    <edition/>
    <issue/>
    <volume/>
    <type>conferenceobject</type>
    <publisherName/>
    <publisherPlace/>
    <creatingCorporation/>
    <contributingCorporation/>
    <belongsToBibliography>0</belongsToBibliography>
    <completedDate>--</completedDate>
    <publishedDate>--</publishedDate>
    <thesisDateAccepted>--</thesisDateAccepted>
    <title language="eng">PYTHIA - AI SUGGESTED INDIVIDUAL LEARNING PATHS FOR EVERY STUDENT</title>
    <abstract language="eng">During the COVID-19 pandemic, the importance of digital course rooms, where teachers provide their learning materials, increased dramatically. While these platforms are crucial for providing teaching materials, they often fall short in addressing individual student needs. A system within an academic setting, capable of creating and presenting individual learning paths for each student, can solve these issues. These paths are composed of various learning elements - defined in our previous work as units of educational content with which a learner works.&#13;
&#13;
Currently, there is no suitable system that enables the integration of learning path generating algorithms into a digital course room. Therefore we present an application that enables this integration into the Moodle Learning Management System (LMS). More precisely, this paper presents a Moodle plugin together with its framework. It describes the mechanism for effectively collecting data from Moodle, which AI algorithms then use to generate personalized learning paths. Subsequently these paths are visualized with the help of the Moodle plugin.&#13;
&#13;
We started with a set of requirements and use cases for the interface connecting Moodle to the AI system, which were established with a group of experts. Based on the requirements, various relevant technologies were assessed, and the best ones were chosen for implementation.&#13;
&#13;
Following that, the paper develops a strategy for software structuring as well as an architecture, focusing on performance, modularity, and ease of deployment for widespread use. Furthermore, the architecture ensures a simple method for integrating the algorithms. Afterwards, the framework's concrete implementation is described. A technique for enriching learning elements with metadata is presented, and additionally a concept for presenting these learning elements within a hierarchy. Moreover, it is shown how questionnaire responses and learning analytics are utilized for data collection. We cover in detail techniques for extracting and storing data from the Moodle database, as well as methods for customizing Moodle course rooms and a standard API for incorporating AI algorithms.&#13;
&#13;
Finally, the paper discusses the application of the proposed framework in an actual course and how student feedback is collected, which could enhance the framework. It concludes with an assessment of the outcomes obtained and prospects for the framework's future advancements.</abstract>
    <parentTitle language="eng">INTED2024 Proceedings: 18th International Technology, Education and Development Conference, Valencia, Spain. 4-6 March, 2024</parentTitle>
    <identifier type="doi">10.21125/inted.2024.0783</identifier>
    <identifier type="isbn">978-84-09-59215-9</identifier>
    <identifier type="issn">2340-1079</identifier>
    <enrichment key="opus.doi.autoCreate">false</enrichment>
    <enrichment key="opus.urn.autoCreate">true</enrichment>
    <licence>Keine Lizenz - Es gilt das deutsche Urheberrecht: § 53 UrhG</licence>
    <author>Simon Röhrl</author>
    <author>Susanne Staufer</author>
    <author>Vamsi Krishna Nadimpalli</author>
    <author>Flemming Bugert</author>
    <author>Flemming Bugert</author>
    <author>Florian Hauser</author>
    <author>Lisa Grabinger</author>
    <author>Dominik Bittner</author>
    <author>Timur Ezer</author>
    <author>Jürgen Mottok</author>
    <subject>
      <language>eng</language>
      <type>uncontrolled</type>
      <value>Personalized Learning Paths</value>
    </subject>
    <subject>
      <language>eng</language>
      <type>uncontrolled</type>
      <value>Learning Management System</value>
    </subject>
    <subject>
      <language>eng</language>
      <type>uncontrolled</type>
      <value>Software Architecture</value>
    </subject>
    <subject>
      <language>eng</language>
      <type>uncontrolled</type>
      <value>Moodle</value>
    </subject>
    <subject>
      <language>eng</language>
      <type>uncontrolled</type>
      <value>Artificial Intelligence</value>
    </subject>
    <collection role="institutes" number="FakEI">Fakultät Elektro- und Informationstechnik</collection>
    <collection role="othforschungsschwerpunkt" number="16311">Digitalisierung</collection>
    <collection role="institutes" number="">Laboratory for Safe and Secure Systems (LAS3)</collection>
  </doc>
</export-example>
