Anton Chagovec
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Anton Chagovec has a total of 6 co-authors. Here are the top co-authors by number of shared publications:
- Antonina Ivanova
Department of Computer Science, Varna Free University “Chernorizets Hrabar”, 84 Yanko Slavchev Str., Chaika Resort, 9007 Varna, Bulgaria
3 shared publications - Fatima Sapundzhi
Department of Communication and Computer Engineering, Faculty of Engineering, South-West University “Neofit Rilski”, 66 Ivan Mihailov Str., 2700 Blagoevgrad, Bulgaria
3 shared publications - Andriana Ivanova
Department of Computer Science, Varna Free University “Chernorizets Hrabar”, 84 Yanko Slavchev Street, Chaika Resort, 9007 Varna, Bulgaria
2 shared publications - Teodora Bakardjieva
Department of Computer Science, Varna Free University “Chernorizets Hrabar”, 9007 Varna, Bulgaria
2 shared publications - Veselina Spasova
Department of Computer Science, Varna Free University “Chernorizets Hrabar”, 84 Yanko Slavchev Street, Chaika Resort, 9007 Varna, Bulgaria
2 shared publications - Meglena Lazarova
Department of Mathematical Modelling and Numerical Methods, Faculty of Applied Mathematics and Informatics, Technical University of Sofia, 8 “Kliment Ohridski” Blvd., 1000 Sofia, Bulgaria
1 shared publication
Latest Publications
- Open Access
- 9 November 2026
- Published by MDPI AG in Engineering Proceedings
- Vol. 167 (1) , 14
A Framework for Interactive Visualization and Evaluation of Classical Shortest-Path Algorithms in a Grid-Based Game
This paper presents the implementation and comparative evaluation of three classical shortest-path algorithms-Dijkstra, A*, and Bellman–Ford-within the Python/Pygame grid-based game environment PetWars. The developed framework provides a common environment for implementation, interactive visualization, and comparative benchmarking of the three algorithms under identical movement rules and map configurations. This common setup enables reproducible evaluation of path-finding behavior. The game models a 14 × 10 grid of walkable and blocked tiles, where movement is allowed in eight directions with weighted orthogonal and diagonal costs. The original PetWars map is used as an interactive demonstration environment. A dedicated demonstration mode visualizes the search process step by step by displaying the explored vertices, the frontier set, and the resulting shortest path. Performance is evaluated using five predefined routes and repeated measurements on two hardware configurations. Execution time, path cost, and the number of expanded vertices are analyzed under identical experimental conditions. The evaluated benchmark results agree with the theoretical characteristics of the evaluated algorithms. Within the implemented framework, A* explores fewer vertices and achieves the shortest execution times owing to heuristic guidance; Dijkstra examines a larger portion of the search space, whereas Bellman–Ford incurs the highest computational cost because of repeated edge relaxation. The developed application provides a common environment for benchmarking and visual demonstration of classical shortest-path algorithms.
- 9 November 2026
- Published by MDPI AG in Engineering Proceedings
- Vol. 167 (1) , 14
Open AccessA Framework for Interactive Visualization and Evaluation of Classical Shortest-Path Algorithms in a Grid-Based Game
This paper presents the implementation and comparative evaluation of three classical shortest-path algorithms-Dijkstra, A*, and Bellman–Ford-within the Python/Pygame grid-based game environment PetWars. The developed framework provides a common environment for implementation, interactive visualization, and comparative benchmarking of the three algorithms under identical movement rules and map configurations. This common setup enables reproducible evaluation of path-finding behavior. The game models a 14 × 10 grid of walkable and blocked tiles, where movement is allowed in eight directions with weighted orthogonal and diagonal costs. The original PetWars map is used as an interactive demonstration environment. A dedicated demonstration mode visualizes the search process step by step by displaying the explored vertices, the frontier set, and the resulting shortest path. Performance is evaluated using five predefined routes and repeated measurements on two hardware configurations. Execution time, path cost, and the number of expanded vertices are analyzed under identical experimental conditions. The evaluated benchmark results agree with the theoretical characteristics of the evaluated algorithms. Within the implemented framework, A* explores fewer vertices and achieves the shortest execution times owing to heuristic guidance; Dijkstra examines a larger portion of the search space, whereas Bellman–Ford incurs the highest computational cost because of repeated edge relaxation. The developed application provides a common environment for benchmarking and visual demonstration of classical shortest-path algorithms.
- Open Access
- 29 June 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (13) , 6454
AI-Driven Threat Detection and Automated Incident Response for Securing Cloud Workloads
The increasing adoption of cloud computing has expanded organizational attack surfaces and created additional opportunities for identity abuse, ransomware operations, data exposure, and configuration-related security incidents. Conventional monitoring environments based primarily on static rules, fragmented telemetry, and manual triage often struggle to prioritize high-severity incidents in real time. This study evaluates the operational impact of an integrated AI-augmented cloud-native SIEM/XDR/SOAR architecture for cloud threat detection and automated incident response. A sequential mixed-methods comparative case study was conducted across two enterprise-style security environments: an AI-augmented architecture combining cloud-native SIEM, XDR telemetry unification, behavioral analytics, AI-assisted correlation, generative-AI analyst support, and SOAR automation, and a conventional baseline environment based on manual triage and signature-based controls. Three attack scenarios were analyzed: phishing-led account takeover, multi-stage ransomware, and shadow-IT data exfiltration. The AI-augmented architecture reduced mean time to triage from 17.4 h in the conventional baseline to 10.7 min and enabled ransomware containment in under five minutes through pre-configured automated response playbooks. The results also showed improved prioritization of high-severity incidents, reduced analyst review burden, and a high automated closure rate. The findings provide operational evidence for the evaluated security architecture. Limitations include single-vendor dependency, non-equivalent false-positive classification mechanisms, proprietary model internals, calibration requirements, and detection gaps involving legitimate third-party services and password-protected content.
- 29 June 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (13) , 6454
Open AccessAI-Driven Threat Detection and Automated Incident Response for Securing Cloud Workloads
The increasing adoption of cloud computing has expanded organizational attack surfaces and created additional opportunities for identity abuse, ransomware operations, data exposure, and configuration-related security incidents. Conventional monitoring environments based primarily on static rules, fragmented telemetry, and manual triage often struggle to prioritize high-severity incidents in real time. This study evaluates the operational impact of an integrated AI-augmented cloud-native SIEM/XDR/SOAR architecture for cloud threat detection and automated incident response. A sequential mixed-methods comparative case study was conducted across two enterprise-style security environments: an AI-augmented architecture combining cloud-native SIEM, XDR telemetry unification, behavioral analytics, AI-assisted correlation, generative-AI analyst support, and SOAR automation, and a conventional baseline environment based on manual triage and signature-based controls. Three attack scenarios were analyzed: phishing-led account takeover, multi-stage ransomware, and shadow-IT data exfiltration. The AI-augmented architecture reduced mean time to triage from 17.4 h in the conventional baseline to 10.7 min and enabled ransomware containment in under five minutes through pre-configured automated response playbooks. The results also showed improved prioritization of high-severity incidents, reduced analyst review burden, and a high automated closure rate. The findings provide operational evidence for the evaluated security architecture. Limitations include single-vendor dependency, non-equivalent false-positive classification mechanisms, proprietary model internals, calibration requirements, and detection gaps involving legitimate third-party services and password-protected content.
- Open Access
- 20 May 2026
- Published by MDPI AG in Preprints.org
AI‑Driven Threat Detection and Automated Incident Response for Securing Cloud Workloads
The increasing adoption of cloud computing has expanded organizational attack surfaces and increased exposure to credential compromise, ransomware, and cloud misconfiguration. Conventional security information and event management (SIEM) systems, based primarily on static correlation rules and signature-based detection, often struggle to process heterogeneous cloud telemetry and prioritize high-severity incidents in real time. This study evaluates the operational impact of an artificial intelligence (AI)-augmented SIEM and Extended Detection and Response (XDR) architecture for cloud threat detection and automated incident response. A mixed-methods comparative case study was conducted across two enterprise-style security environments: an AI-augmented cloud-native SIEM/XDR architecture and a conventional baseline environment based on manual triage and signature-based controls. Three attack scenarios were analyzed: phishing-led account takeover, multi-stage ransomware, and shadow-IT data exfiltration. The AI-augmented environment reduced mean time to triage from 17.4 hours in the conventional baseline to 10.7 minutes and enabled ransomware containment in under five minutes through automated response playbooks. The results also showed improved prioritization of high-severity incidents, reduced analyst workload, and a high automated closure rate. However, limitations were observed in the calibration of behavioral models, vendor dependency, and detection gaps involving legitimate third-party services and password-protected content. The findings should be interpreted as operational evidence at an architectural level, not as an isolated evaluation of individual AI models.
- 20 May 2026
- Published by MDPI AG in Preprints.org
Open AccessAI‑Driven Threat Detection and Automated Incident Response for Securing Cloud Workloads
The increasing adoption of cloud computing has expanded organizational attack surfaces and increased exposure to credential compromise, ransomware, and cloud misconfiguration. Conventional security information and event management (SIEM) systems, based primarily on static correlation rules and signature-based detection, often struggle to process heterogeneous cloud telemetry and prioritize high-severity incidents in real time. This study evaluates the operational impact of an artificial intelligence (AI)-augmented SIEM and Extended Detection and Response (XDR) architecture for cloud threat detection and automated incident response. A mixed-methods comparative case study was conducted across two enterprise-style security environments: an AI-augmented cloud-native SIEM/XDR architecture and a conventional baseline environment based on manual triage and signature-based controls. Three attack scenarios were analyzed: phishing-led account takeover, multi-stage ransomware, and shadow-IT data exfiltration. The AI-augmented environment reduced mean time to triage from 17.4 hours in the conventional baseline to 10.7 minutes and enabled ransomware containment in under five minutes through automated response playbooks. The results also showed improved prioritization of high-severity incidents, reduced analyst workload, and a high automated closure rate. However, limitations were observed in the calibration of behavioral models, vendor dependency, and detection gaps involving legitimate third-party services and password-protected content. The findings should be interpreted as operational evidence at an architectural level, not as an isolated evaluation of individual AI models.