Hochschule Mittweida

We present a novel approach for assessing the reliability of error-prone systems and networks using an interpretable prototype-based classification framework. Specifically, reliability levels of consecutive k-out-of-n success systems, failure networks, and domination networks are classified using Generalized Matrix Learning Vector Quantization (GMLVQ). Beyond achieving accurate classification, the proposed method provides informative insights into the influence of input probabilities on the resulting reliability levels. To further enhance the interpretability of the learned relevance matrix in GMLVQ, we introduce two graph-based visualization strategies that reveal structural patterns and relevant feature interactions that are critical for reliability-level classification. The proposed approach is generally applicable to arbitrary coherent systems and can be further adapted to estimate the probability of a union of any finite family of events based on their individual and pairwise intersection probabilities.

The superposition principle in quantum mechanics enables the encoding of an entire solution space within a single quantum state. By employing quantum routines such as amplitude amplification or the Quantum Approximate Optimization Algorithm (QAOA), this solution space can be explored in a computationally efficient manner to identify optimal or near-optimal solutions. In this article, we propose quantum circuits that operate on binary data representations to address a central task in prototype-based classification and representation learning, namely the so-called winner determination, which realizes the nearest prototype principle. We investigate quantum search algorithms to identify the closest prototype during prediction, as well as quantum optimization schemes for prototype selection in the training phase. For these algorithms, we design oracles based on arithmetic circuits that leverage quantum parallelism to apply mathematical operations simultaneously to multiple inputs. Furthermore, we introduce an oracle for prototype selection, integrated into a learning routine, which obviates the need for formulating the task as a binary optimization problem and thereby reduces the number of required auxiliary variables. All proposed oracles are implemented using the Python 3-based quantum machine learning framework PennyLane and empirically validated on synthetic benchmark datasets.

Defacing of brain magnetic resonance imaging (MRI) scans by removing identifiable facial features is essential for protecting patient privacy, yet assessing defacing quality remains challenging. While deep learning methods offer solutions, they require large labeled datasets, limiting their practical applicability. This study presents DefaceQA, a machine learning (ML) approach for automated defacing quality assessment using quantitative image features. A dataset of 200 MRI scans from the Leukodystrophy Registry at the Leipzig University Medical Center was processed using four defacing algorithms: PyDeface, QuickShear, FSL-Deface, and MRI-Deface. Image features extracted from original and defaced scans were used to classify defacing efficacy. The ML classifiers achieved an AUROC of 0.84 and an accuracy of 0.85 under a lenient criterion for successful/unsuccessful defacing, with the Feature Similarity Index Measure (FSIM) emerging as a key predictor. The findings demonstrate ML’s potential for defacing evaluation while highlighting challenges related to dataset limitations and generalizability.

Integrative biomolecular modeling uses experimental data to validate computationally generated structures. We use Förster-Resonance-Energy transfer (FRET) as an experimental distance measure on the molecular level to evaluate de novo predicted ribonucleic acid (RNA) 3D structure collections. FRET-assisted RNA structure prediction requires an intricate interplay of multiple software components. It involves generating an initial RNA 3D model, performing in silico fluorophore labeling, and computing FRET observables from molecular dynamics (MD) trajectories. To automate this workflow, we developed a Python-based computational pipeline that integrates FRET as an experimental constraint for RNA structure validation.

In the practical work of the tool- and die-making industry, the transformation process frequently involves the manual reworking of certain pieces and surfaces. Therefore, the quality of the final product depends on the expertise of the toolmaker. The objective of this research is to ease the transfer of knowledge by enabling the virtual visualization of tryout tools for novices and to alleviate the process of further education. The employment of an augmented reality-based assistant is expected to facilitate the comprehension of tryout tools and stamping dies by new employees. Due to interviews with experts in the toolmaking industry, it has been revealed that this approach could enhance the familiarization process prior to hands-on training.