Repository der Technischen Hochschule Ingolstadt
Not a member yet
4170 research outputs found
Sort by
SEGym: Optimizing Large Language Model Assisted Software Engineering Agents with Reinforcement Learning
No full textAn interdisciplinary perspective on AI-supported decision making in medicine
Get PDFArtificial intelligence (AI)-supported medical diagnosis offers the potential to utilize the collaborative intelligence of context-sensitive humans and narrowly focused machines for patients’ benefit. The employment of machine-learning-based decision-support systems (MLDSS) in medicine, however, raises important multidisciplinary challenges that cannot be addressed in isolation. We discuss three disciplinary perspectives on the topic and their interplay. Ethical issues arise at the level of changing responsibility structures in healthcare. Behavioral issues relate to the actual impact that the system has on physicians. Technical issues arise with respect to the training of a machine learning (ML) model that gives accurate advice. We argue that the interaction between physicians and MLDSS including the concrete design of the interface in which this interaction occurs can only be considered at the intersection of all three disciplines
Evaluation of Automotive Radar Simulation Tools for Adverse Weather Conditions: A Comparative Analysis of Real Measurement, HFSS and Monte-Carlo GO Methods
No full textBenchmarking of Retrieval Augmented Generation: A Comprehensive Systematic Literature Review on Evaluation Dimensions, Evaluation Metrics and Datasets
Get PDFDespite the rapid advancements in the field of Large Language Models (LLM), traditional benchmarks have proven to be inadequate for assessing the performance of Retrieval Augmented Generation (RAG) systems. Therefore, this paper presents a comprehensive systematic literature review of evaluation dimensions, metrics, and datasets for RAG systems. This review identifies key evaluation dimensions such as context relevance, faithfulness, answer relevance, correctness, and citation quality. For each evaluation dimension, several metrics and evaluators are proposed on how to assess them. This paper synthesizes the findings from 12 relevant papers and presents a concept matrix that categorizes each evaluation approach. The results provide a foundation for the development of robust evaluation frameworks and suitable datasets that are essential for the effective implementation and deployment of RAG systems in real-world applications
Pattern-based parsing of German traffic regulations (StVO) for legal knowledge graph construction
No full text3D‐Printing with steel of a bolted connection
Get PDFAdditive Manufacturing seems promising for the steel construction industry, especially for small components with complex shapes. With Wire‐and‐Arc‐Additive‐Manufacturing (WAAM) a process was found which is fast and cheap compared to other Additive Manufacturing processes for metals. Furthermore, arc‐welding is well known in steel construction industry.Bolted head plates are a common connection. Although they are relatively easy to produce, they are uneconomical in load transfer due to straining of the plates which are subject to bending. This type of connection is predestined for Additive Manufacturing. Rethinking the structure by taking into account the potentials of Additive Manufacturing brings a clear advantage compared to the conventional production. Thus, with a material saving of 60 % the same load capacity of the original flat head plate can be achieved.The paper shows how the new production process WAAM can significantly change the shape of structures on the example of a bolted head plate. Therefore, all steps from the original design to the manufactured structure are demonstrated, including the structure finding by using topology optimization, the manufacturing with different parameters and a first testing of the manufactured structures