arXiv subjects

Lukas Bahr

Publications and source records attributed to Lukas Bahr.

4 recordsLinked to original sources

arXiv2025

Sensitivity analysis of image classification models using generalized polynomial chaos

Integrating advanced communication protocols in production has accelerated the adoption of data-driven predictive quality methods, notably machine learning (ML) models. However, ML models in image classification often face significant uncertainties arising from model, data, and domain shifts. These uncertainties lead to overconfidence in the classification model's output. To better understand these models, sensitivity analysis can help to analyze the relative influence of input parameters on the output. This work investigates the sensitivity of image classification models used for predictive quality. We propose modeling the distributional domain shifts of inputs with random variables and quantifying their impact on the model's outputs using Sobol indices computed via generalized polynomial chaos (GPC). This approach is validated through a case study involving a welding defect classification problem, utilizing a fine-tuned ResNet18 model and an emblem classification model used in BMW Group production facilities.

cs.LG↗

arXiv2024

Anwendung von Causal-Discovery-Algorithmen zur Root-Cause-Analyse in der Fahrzeugmontage

Root Cause Analysis (RCA) is a quality management method that aims to systematically investigate and identify the cause-and-effect relationships of problems and their underlying causes. Traditional methods are based on the analysis of problems by subject matter experts. In modern production processes, large amounts of data are collected. For this reason, increasingly computer-aided and data-driven methods are used for RCA. One of these methods are Causal Discovery Algorithms (CDA). This publication demonstrates the application of CDA on data from the assembly of a leading automotive manufacturer. The algorithms used learn the causal structure between the characteristics of the manufactured vehicles, the ergonomics and the temporal scope of the involved assembly processes, and quality-relevant product features based on representative data. This publication compares various CDAs in terms of their suitability in the context of quality management. For this purpose, the causal structures learned by the algorithms as well as their runtime are compared. This publication provides a contribution to quality management and demonstrates how CDAs can be used for RCA in assembly processes.

cs.LG↗

arXiv2024

Knowledge graph enhanced retrieval-augmented generation for failure mode and effects analysis

Failure mode and effects analysis (FMEA) is an essential tool for mitigating potential failures, particularly during the ramp-up phases of new products. However, its effectiveness is often limited by the reasoning capabilities of the FMEA tools, which are usually tabular structured. Meanwhile, large language models (LLMs) offer novel prospects for advanced natural language processing tasks. However, LLMs face challenges in tasks that require factual knowledge, a gap that retrieval-augmented generation (RAG) approaches aim to fill. RAG retrieves information from a non-parametric data store and uses a language model to generate responses. Building on this concept, we propose to enhance the non-parametric data store with a knowledge graph (KG). By integrating a KG into the RAG framework, we aim to leverage analytical and semantic question-answering capabilities for FMEA data. This paper contributes by presenting set-theoretic standardization and a schema for FMEA data, an algorithm for creating vector embeddings from the FMEA-KG, and a KG-enhanced RAG framework. Our approach is validated through a user experience design study, and we measure the precision and performance of the context retrieval recall.

cs.IR↗

arXiv2023

Towards gain tuning for numerical KKL observers

This paper presents a first step towards tuning observers for general nonlinear systems. Relying on recent results around Kazantzis-Kravaris/Luenberger (KKL) observers, we propose an empirical criterion to guide the calibration of the observer, by trading off transient performance and sensitivity to measurement noise. We parametrize the gain matrix and evaluate this criterion over a family of observers for different parameter values. We then use neural networks to learn the mapping between the observer and the nonlinear system, and present a novel method to sample the state-space efficiently for nonlinear regression. We illustrate the merits of this approach in numerical simulations.

eess.SY↗