New directions in scientific research in innovative interdisciplinary solutions International Interdisciplinary PhD Workshop 2025 (ebook) Bochnia

This publication was prepared as a permanent record of research findingspresented by doctoral students and academics representing diverse researchbackgrounds and schools. The focus is on interdisciplinary solutions, combiningengineering, computer science, and mathematical methods with approaches relevantto management science and socio-economic applications. The International Interdisciplinary PhD Workshop (IIPhDW) is a cyclicaland international conference, and its primary function is to provide a platform forresearch presentations, knowledge exchange, and collaboration between youngscientists from various disciplines. The 2025 edition featured a particularly strongtechnical component, encompassing artificial intelligence, computer science, automationand control, robotics and mechatronics, telecommunications, signalprocessing, as well as mechanical and production engineering. At the same time,the inclusion of topics in economics and management confirmed the workshop'sbroad scope and its ability to integrate research perspectives relevant to contemporarytechnological and organizational challenges. A significant group of publications includes works on process and biomedicaltomography, as well as image reconstruction methods using machine learningand deep learning, including approaches combining physical models with neuralnetwork architectures. Examples include research on image reconstruction inelectrical impedance tomography, the integration of tomography with neural networksfor industrial process monitoring, and the use of ultrasound tomographyin measurement and reconstruction analysis. This theme is further reinforced byworks on hybrid tomography systems, process monitoring using mixed realitytechnology, and applications in the areas of physiological parameter monitoringand non-invasive diagnostics. The second recognizable axis is artificial intelligence in IT and cyberphysicalsystems, encompassing both the construction of predictive and classificationmodels and their implementation in industrial, medical, and service environments.This trend includes work related to the application of machine learningmethods in system and network security, solutions based on LLM agents inproject team workflows, processing unstructured data using OCR and languagemodels, and multimodal analysis in intelligent customer service systems. Thisperspective highlights the contemporary trend of convergence of AI techniqueswith data engineering, software engineering, and systems integration, which hasdirect implications for the design of scalable implementation solutions. The third thematic area covers embedded systems, communication, and signalprocessing, along with elements of computational resource optimization. Themonograph includes papers on, among other things, phase shift estimation innoisy environments, pseudorandom sequence generation, acoustic feature detection,and the efficiency of machine learning applications at the network edge inthe context of Kubernetes scheduling heuristics. In the monograph, this strandserves a methodological purpose, providing signal analysis tools and computationalmechanisms that form the foundation for many AI and measurement systemapplications. A significant complement to the technical perspective are works in the areasof management and organizational and economic analysis, which address theneed to understand the determinants of technology implementation and the conductof innovative projects. The publications address, among other things, thepredictors of success in startup management and the analysis of organizationalimprovements in public institutions. Their presence strengthens the interdisciplinarynature of the monograph by demonstrating that the effectiveness of engineeringsolutions depends not only on the quality of algorithms and devices, butalso on the organizational, process, and decision-making context. The monograph is intended as a reference for academics and doctoral students,particularly those seeking examples of research that combines theory withapplication. The collected papers offer a comprehensive overview of research activitiestypical of early careers in science, from conceptual studies and methodanalysis, through device and software architecture prototyping, to experimentsand evaluation of the effectiveness of proposed solutions. At the same time, thepublication allows for the identification of common methodological denominators,such as the growing importance of measurement data, simulation, deeplearning, systems integration, and the pursuit of real-time operation in industrialand biomedical environments. The introduction, on the one hand, contextualizes the monograph within themission of IIPhDW as a workshop supporting researcher development and theinternationalization of research. On the other hand, it organizes the chapter topicsin the perspective of dominant technological trends and application needs thatpermeate various fields. Consequently, the monograph can be viewed as a syntheticoverview of current research directions for doctoral students and youngacademics, as well as an inspiration for undertaking work combining artificialintelligence methods, measurement systems, software engineering, and managementanalyses within modern interdisciplinary projects. Spis treści: Introduction ... 14Monika Kulisz, Grzegorz Kłosowski, Tomasz RymarczykPerformance analysis of the differential neural network architecture in industrial tomography ... 16Introduction ... 161. Methods ... 172. Results ... 193. Discussion and Conclusions ... 21References ... 22 Grzegorz Kłosowski, Tomasz Rymarczyk, Konrad NiderlaApplication of Multi-Head Neural Network Structures in Process Tomography ... 24Introduction ... 241. Methods ... 252. Results ... 273. Discussion and Conclusions ... 29References ... 30 Dariusz MajerekPINN-VIT: a physics-informed neural network enhanced with vision transformer for image reconstruction in electrical impedance tomography ... 32Introduction ... 321. Methods ... 332. Results ... 353. Discussion and Conclusions ... 36References ... 38 Barbara Stefaniak, Dariusz Wójcik, Tomasz RymarczykGenerate CT pictures for human torse using AI ... 40Introduction ... 401. Motivation ... 412. Methodology ... 412.1. Dataset preprocessing ... 412.2. Adapted CrossViT Architecture for 3D CT Generation ... 412.3. Decoder and training objective ... 423. Results ... 434. Conclusion ... 43References ... 45 Oleksii Hyka, Dariusz Wójcik, Tomasz RymarczykModular Software Architecture of the Portable Defektoskop Device ... 46Introduction .... 461. System Architecture ... 462. Core Services ... 473. Update Management ... 484. Communication Framework ... 485. 3D Visualization System ... 496. Results ... 497. Conclusion ... 50References ... 51 Marcin Dziadosz, Tomasz Rymarczyk, Dariusz MajerekApplication of machine learning algorithms in measurement and reconstruction analysis using Ultrasound Tomography ... 52Introduction ... 521. Transmission ultrasound tomography .... 532. Methods ... 533. Transmission UST forward problem ... 544. Transmission UST inverse problem ... 555. Conclusion ... 58References ... 59 Yadu Krishnan Krishnakumar, Andreas Ahrens, Christoph Lange, Jelena Zaščerinska, Olaf GroteMeasuring burstiness in randomised bit sequences ... 60Introduction ... 621. Model Basics ... 622. Measuring Burstiness ... 633. Practical Application ... 644. Comparison with NIST Test Results ... 655. Conclusions ... 65References ... 66 Grzegorz Rybak, Dariusz Wójcik, Tomasz RymarczykPerformance Analysis of Holography and Mixed Reality Systems for Industrial Process Tomography SUPERVISION ... 68Introduction ... 681. Mixed reality for IPT ... 692. Algorithms and methods ... 703. Results and discussion ... 724. Conclusions and future works ... 77References .... 78Jacek KorzeniakPredictors of Success in Startup Management ... 80Introduction ... 801. Methodology ... 812. Challenges in the context of the startup definition ... 813. Defining startup success criteria ... 824. Characteristics of a Startup Founder ... 835. Conclusions .... 85References ... 85 Sabrina Ferdous, Radosław WajmanPersonalized Modeling and Deep Learning–Based Temporal Prediction of Pediatric Bladder Behavior .... 86Introduction ... 861. Related Work ... 872. Methodology .... 882.1. Personalized Bladder Simulation ... 882.2. Proposed Deep Learning Method ... 882.3. Training Details ... 892.4. Datasets and Setup ... 893. Results ... 904. Discussion and Future Work ... 905. Conclusions ... 91References ... 91Andreas Wenzel, Andreas Ahrens, Yadu Krishnan Krishnakumar, Ingo MüllerGenerating randomised bit sequences using sigma-delta converter ... 94Introduction ... 941. Model Basics .... 962. First Results .... 99References ... 99Jan Bartelt, Olaf HagendorfStudy of anomaly detection methods in unobtrusively acquired data in the context of AAL ... 102Introduction .... 1021. Motivation .... 1032. Available Data .... 1033. Application Requirements ... 1054. Algorithm Requirements ... 1055. Algorithm Selection ... 1066. Conclusion ... 106References .... 107Dominik Gnaś, Tomasz Rymarczyk, Michał Oleszek, Dariusz WójcikDevelopment of a Hybrid Non-Invasive Glucose Sensor Combining Electrical Impedance and Optical Spectroscopy ... 110Introduction .... 1101. Discussion of measurement methods .... 1112. Development of the hardware layer .... 1113. Conclusion .... 113References .... 115Michał Styła, Przemysław Adamkiewicz, Tomasz RymarczykRadar-based location system using high-frequency signals and reflective microwave tomography for indoor human detection and positioning ... 116Introduction ... 1161. System structure and hardware layer .... 1172. Data acquisition and processing .... 118References .... 121 Patryk Marek, Alicja Rachwał, Jakub Pizoń, Nina Krawczak, Paweł WoźniakAutomated Project Team Design Using LLM-Based Agents ... 122Introduction ... 1221. Proposed team assembly system .... 1232. Case study .... 1253. Conclusions and future work ... 126References .... 127Alicja Rachwał, Jakub Pizoń, Nina Krawczak, Patryk Marek, Paweł WoźniakPerformance Testing with Locust: A Flexible Approach to Load Simulation ... 128Introduction ... 1281. Locust description ... 1292. Case study ... 1303. Conclusions .... 132References ... 133Michał Styła, Dariusz Wójcik, Przemysław AdamkiewiczImplementation of a miniature measurement platform for asset location using ultra-wideband signals and time-of-flight distance measurement methods ... 136Introduction ... 1361. Development of the hardware layer .... 1372. Localization algorithms and noise reduction ... 1383. Conclusion ... 142References ... 143 Marek WójcikIntroduction to research on using artificial intelligence for automation and optimisation of it system and computer network security ... 144Introduction ... 1441. Scope and Thesis .... 1452. Architecture Overview .... 1453. Scenario Catalogue .... 1464. Data Strategy and Governance .... 1475. Numerical Modelling .... 1476. Text Understanding with LLMs .... 1487. Fusion Policies .... 1488. Evaluation Methodology .... 1499. Human Factors, Ethics, and Interpretability .... 14910. Generalisability and Expected Outcomes .... 149References .... 150 Łukasz GugałaDetecting sound features using python .... 152Introduction ..... 1521. Why use Python for detecting features? ... 1532. Machine learning based on audio features .... 1533. Normalization and noise removing .... 1544. How to train models? .... 155References .... 158 Mateusz WielebaAreas for improvement in the police force and economic crime ... 160Introduction ... 1601. Current Challenges in Combating Economic Crime .... 1602. Police Education System and the Need for Reforms .... 1623. Priority Areas for Improvement .... 1634. Conclusion .... 164References .... 165 Michał Gołąbek, Barbara Stefaniak, Tomasz Rymarczyk, Dariusz WójcikEnhancing Geometric Accuracy of 3D Ultrasonic Imaging: A Weighted Reconstruction Approach for Nondestructive Testing ... 166Introduction ... 1661. Aims and novelties ... 1672. Hardware ... 1673. Methods .... 1694. Results and conclusions ... 170References ... 171 Michał Maj, Kamil Krawczak, Szymon Olędzki, Damian Pliszczuk, Tomasz Rymarczyk, Tomasz CieplakSemantic Extraction of Medical Data from Unstructured Documents Using OCR, LLM, and GraphQL .... 174Introduction ... 1741. OCR and Semantic Extraction ... 1753. GraphQL Layer and Semantic Interface .... 1774. Conclusions .... 180References .... 180 Michał Maj, Łukasz Maciura, Jakub Pizoń, Damian Pliszczuk, Tomasz Rymarczyk, Tomasz CieplakCross-Modal Computer Vision and Emotional Recognition in Intelligent Customer Service Systems ... 182Introduction ... 1821. Related Work .... 1833. Dataset and Methodology ... 1864. Experiments and Results .... 1875. Conclusion and Future Work ... 188References .... 189Krzysztof Król, Monika Kulisz, Grzegorz Kłosowski, Tomasz RymarczykApplication of electrical impedance tomography to monitor the alcoholic fermentation process in real time ... 190Introduction ... 1901. Materials and Methods ... 1912. Results and Discussion .... 193References .... 195Tomasz Rymarczyk, Grzegorz Kłosowski, Monika Kulisz, Konrad NiderlaCRYSTALLIZATION control environment using hybrid tomography and deep reinforcement learning ... 198Introduction .... 1981. Model Components ... 2002. Reinforcement Learning Methods .... 2013. Conclusions .... 202References .... 202Michał Oleszek, Tomasz ŁobodiukEfficient Phase Shift Estimation in Noisy Discrete-Time Signal Processing ... 204Introduction ... 2041. Methodology ... 2042. Algoritm .... 2042.1 Discrete Fourier Transform (DFT) ... 2052.2 Curve Fit ... 2052.3 Least Squares (LSQ) ... 2052.4 Hilbert Transform .... 2052.5 Quadrature Demodulation ... 2062.6 Cross-Correlation .... 2063. Results .... 2064. Conclusion .... 210References ... 211 Michał Gołąbek, Dariusz Majerek, Tomasz Rymarczyk, Krzysztof KrólIntegration of Reconstruction and Machine Learning Methodsin Industrial Process Monitoring ... 212Introduction .... 2121. Hardware Construction .... 2132. Methods ... 2143. Results .... 215References .... 217Michał Gołąbek, Grzegorz Kłosowski, Marcin Dziadosz, Dariusz Wójcik, Tomasz RymarczykDevelopment of a portable, bimodal tomographic system for monitoring the lower urinary tract .... 218Introduction .... 2181. Hardware Construction ... 220References ..... 222Paweł KaletaEfficient cloud resource management .... 224Introduction .... 2241. Cloud Models ... 2242. Resource Management Strategies .... 2253. Platforms and Tools .... 2264. Security Aspects ... 2275. Energy Efficiency and Costs ... 2286. Conclusion ... 228References .... 229Krzysztof Król, Grzegorz Kłosowski, Monika Kulisz, Tomasz RymarczykIntegration of Hybrid Tomography and Neural Networks in Industrial Process Monitoring ... 230Introduction .... 2301. Materials and Methods .... 2312. Results and discussion .... 232References .... 234Monika Kulisz, Grzegorz Kłosowski, Tomasz Rymarczyk, Konrad NiderlaApplication of electrical impedance tomography to real-time monitoring and optimization processes .... 2361. Fermentation Process .... 2372. Methods ... 237References ... 241Łukasz Maciura, Krzysztof Król, Grzegorz Rybak, Tomasz RymarczykEvolutionary training of recurrent networks for classification and prediction in industry .... 242Introduction ... 2422. Results and discussion .... 246References ... 247Tomasz Rymarczyk, Marcin Dziadosz, Mariusz Mazurek, Amelia Kosior-Romanowska, Dariusz Wójcik, Krzysztof KrólApplication of Deep Residual Neural Networks for Chemical Compound Identification Based on FTIR Spectra in an Optical Tomography System ... 250Introduction ... 2501. The system .... 2512. The machine learning approach ... 2513. Conclusions ... 255References .... 255Tomasz Rymarczyk, Mariusz Mazurek, Marcin Dziadosz, Konrad NiderlaUsing machine learning and mobile eit sensors for non-invasive urinary tract monitoring .... 258Introduction ... 2581. The method ... 2592. The results .... 2603. Conclusions .... 263References ... 264Paweł BarwiakUsing financial statements to build a picture of how a company operates .... 266Introduction ... 2661. Objective ... 2672. Method ... 2673. Results .... 2684. Conclusion ... 270References .... 271