arXiv subjects

Marc Kurz

Publications and source records attributed to Marc Kurz.

3 recordsLinked to original sources

arXiv2026

Continuous Face Authentication on Mobile and Desktop Platforms: A Comparative Study

Personal devices hold sensitive data and provide access to sensitive services. Conventional personal device authentication verifies users' identity only at the moment access is granted. An unlocked device may be accessed by an unauthorized person if the user stops using the device without locking it, or if another person takes over. Continuous authentication addresses this gap. This paper investigates how device type and usage conditions influence continuous mobile face authentication with an InsightFace-based approach with temporal trust decay. We evaluate the approach with mobile and desktop recordings with different head directions and lighting conditions. We also evaluate recordings from everyday mobile device use without predefined tasks. The results show that device type alone has little impact, while different usage conditions do have impact on the authentication performance. Results also show that everyday mobile device use is in general more challenging for continuous face authentication, where reduced face visibility, including occlusions and faces outside the camera viewport, is a main contributor to false rejections.

cs.HC↗

arXiv2026

Taming the Search Space: Solving and Generating Hitori and Binairo Puzzles

This paper investigates solving and generation techniques for the logic puzzles Hitori and Binairo. Two solving paradigms are compared: backtracking with domain-specific optimizations, and SAT-based solving via conjunctive normal form encodings. An empirical evaluation analyzes runtime, explored search nodes, and branching factor across varying puzzle sizes. To support systematic benchmarking in the evaluation, generators capable of producing valid and uniquely solvable puzzle instances are developed. Results indicate that constraint propagation is the most effective backtracking optimization, substantially reducing the effective branching factor, search tree size, and thus runtime. Heuristic variable ordering and scoring strategies provide additional improvements. For Binairo, the SAT-based approach solves all evaluated instances within low runtime, while optimized backtracking fails to solve difficult puzzle instances within the timeout. For Hitori, propagation-based backtracking achieves the best results, while for the SAT-based approach the iterative connectivity check takes up the majority of the runtime, failing difficult puzzle instances.

cs.LO↗

arXiv2020

Occams Razor for Big Data? On Detecting Quality in Large Unstructured Datasets

Detecting quality in large unstructured datasets requires capacities far beyond the limits of human perception and communicability and, as a result, there is an emerging trend towards increasingly complex analytic solutions in data science to cope with this problem. This new trend towards analytic complexity represents a severe challenge for the principle of parsimony or Occams Razor in science. This review article combines insight from various domains such as physics, computational science, data engineering, and cognitive science to review the specific properties of big data. Problems for detecting data quality without losing the principle of parsimony are then highlighted on the basis of specific examples. Computational building block approaches for data clustering can help to deal with large unstructured datasets in minimized computation time, and meaning can be extracted rapidly from large sets of unstructured image or video data parsimoniously through relatively simple unsupervised machine learning algorithms. Why we still massively lack in expertise for exploiting big data wisely to extract relevant information for specific tasks, recognize patterns, generate new information, or store and further process large amounts of sensor data is then reviewed; examples illustrating why we need subjective views and pragmatic methods to analyze big data contents are brought forward. The review concludes on how cultural differences between East and West are likely to affect the course of big data analytics, and the development of increasingly autonomous artificial intelligence aimed at coping with the big data deluge in the near future.

cs.DB↗