CONTENTS
|
Machine Learning for Football Match-Point Prediction. Algorithms
Performance on Small Datasets Marin FOTACHE, Irina COJOCARIU, Armand BERTEA 5 Match-point prediction is an operationally relevant task in football analytics, supporting tactical preparation, squad management, and performance monitoring. Based on a dataset provided by InStat on the results of a struggling Romanian football team, this study proposes a reproducible and auditable decision-support workflow framed as a supervised binary classification problem that estimates the probability of securing at least one point by match end (90 minutes plus stoppage time) using match data. Predictors are engineered to represent three interpretable constructs: average team age, tactical deployment, and key-player exposure. The workflow was implemented in R using two intertwined frameworks for data processing and exploration (tidyverse) and Machine Learning (tidymodels). Five classification algorithms were benchmarked. Results provide some insights on the classification performance when applied to small datasets. Also, the average team age and key-player minutes emerge as the most important predictors in explaining the variability of point attainment for the reference team. Keywords: Football Analytics, Classification, Match outcome, Applied Machine Learning algorithms, Tidymodels, Feature engineering, Decision support Architecture of a Broadcast and Media Production Learning Center Integrating Cloud-Based Ingest and Automated News Classification Adrian VINTILĂ, Constanța-Nicoleta BODEA 23 This paper presents a reference architecture for a learning center designed to support engineering education in television broadcast and media production. The proposed environment replicates a small-scale but operationally complete television broadcast facility and is intended to give students hands-on access to the equipment, software, and workflows that characterize modern newsrooms. The architecture is organized around four interconnected components that follow the content lifecycle: media acquisition, processing and planning, production, and distribution. Two original subsystems, previously designed and evaluated by the authors, are integrated into this architecture in order to expose students to current technological trends. The first is a cloud-based automated ingest subsystem, embedded within the media acquisition component, which removes manual steps from the file-transfer pipeline and accelerates the availability of mobile journalism media assets. The second is a supervised machine-learning subsystem for the automatic classification of Romanian news stories, embedded within the processing and planning component, which supports editorial organization and newsroom decision-making. Five learning scenarios are defined to illustrate how students engage with the architecture across all four components, ranging from mobile journalism capture and cloud-based ingest to control room operation and on-air delivery through a streaming platform. The paper’s contribution is the consolidated reference architecture and the integration of these two previously evaluated subsystems into a coherent educational framework. A systematic assessment of student learning outcomes is planned for future work. Keywords: Practice-based learning, Media asset management, Hybrid cloud workflow automation, Supervised text classification, Television production Quantum Key Distribution with Enhanced PQC Validation Vlad BERARU, Carmen MILEA 37 Quantum Key Distribution (QKD) provides information-theoretic security based on quantum mechanics, but its practical deployment is limited by its reliance on classical infrastructure and inability to operate as a standalone solution. Recent research emphasizes hybrid approaches that combine QKD with post-quantum cryptography (PQC) to achieve practical and robust security. This paper presents a hybrid key distribution model integrating QKD with the post-quantum key encapsulation mechanism. The two independently generated keys are combined to derive a unified shared secret, enhancing resilience against both classical and quantum adversaries. The system is implemented using a simulated quantum environment, enabling the emulation of quantum communication over classical networks. The proposed approach is validated through a secure image transmission use case, demonstrating the correctness of the hybrid key exchange. The results highlight the feasibility of hybrid classical–quantum cryptographic systems as a practical step toward secure communication in emerging quantum network infrastructures. Keywords: Quantum Key Distribution, Post-Quantum Cryptography, Hybrid Cryptography, Quantum Networks, Secure communication Evaluating the Adversarial Robustness of Deepfake Detectors Mihai-George STURZA, Daria-Maria PREDA 49 As eKYC pipelines see increasingly more usage in verifying identities across banking, insurance and retail applications, bad actors are developing new ways of bypassing validation and gaining trusted access in protected environments. Deepfake detectors have their unique place in such verification pipelines and act as a defense against synthetic forgeries delivered through injection attacks. This paper evaluates the adversarial robustness of four architecturally diverse detectors under white-box and black-box attacks in a cross-domain setting. Under white-box conditions all four detectors are fully compromised, even at perturbation magnitudes that remain metrically imperceptible. Transfer attacks show that adversarial examples crafted against a single, freely available, model can evade other detectors with high reliability and query-based attacks achieve comparable results without any knowledge of model internals. These results indicate that evaluated deepfake detectors do not withstand adversarial manipulation under realistic attack conditions, raising practical concerns for production eKYC deployments and for compliance with the EU AI Act’s robustness requirement for high-risk biometric systems. Keywords: Deepfake detection, eKYC, Biometric security, Adversarial testing Artificial Intelligence as a Co-Creator in Software Development Sabin-Marian ARSENE 60 This paper explores the paradigm shift in software engineering by positioning Artificial Intelli-gence (AI) as an active co-creator within the Software Development Life Cycle (SDLC). While traditional AI tools function as passive assistants, the proposed framework integrates large language models and predictive analytics to foster a collaborative environment between human developers and machine intelligence. Through a Retrieval-Augmented Generation (RAG) approach, the study evaluates improvements in requirements analysis, code generation, and proactive defect detection. Experimental simulations indicate a 57% reduction in lead time and a 37% decrease in defect density, highlighting significant productivity gains. However, the re-search also identifies the "Reviewer’s Paradox," where the cognitive load shifts toward verification and architectural oversight. The findings suggest that AI-human co-creation not only optimizes resource allocation but also necessitates a fundamental redefinition of developer roles in the era of AI-native software engineering. Keywords: Artificial Intelligence, Software development, AI-augmented development, Co-creation, Adaptive software Post-Quantum Cryptographic Approach for Linux Kernel Module Used in Intrusion Detection System Mariuca PRICOP, Cristian TOMA 71 As quantum computers get closer to being used in real-life situations, existing methods to protect against cyber threats are becoming less effective. This paper presents a hybrid, quantum-enhanced intrusion detection system implemented as a Linux kernel module, integrating kernel-level monitoring, user-space intelligence, quantum machine learning (QML) and post-quantum cryptographic (PQC) signature verification. The system uses a zero-trust approach by requiring cryptographic authorization for legitimate actions while simultaneously detecting anomalous behavior. Final enforcement decisions are executed within the kernel, ensuring low-latency response and strong security guarantees. By decoupling heavy cryptographic operations from kernel space and leveraging quantum-enhanced analysis techniques, this approach achieves a balance between performance, scalability, and resilience against future quantum threats. Keywords: Intrusion Detection System, Linux kernel module, Post-Quantum Crypto Publishing Guide for Authors 82 INFOREC Association 84 |
