Digital security plays an ever-increasing, crucial role in today’s information-based society. The variety of threats and attack patterns has dramatically increased with the advent of digital transformation in our lives. Researchers in both public and private sectors have tried to identify new means to counteract these threats, seeking out-of-the-box ideas and novel approaches. Amongst these, data analytics and artificial intelligence/machine learning tools seem to gain new ground in digital defence. However, such instruments are used mainly offline with the purpose of auditing existing IDS/IDPS solutions. We submit a novel concept for integrating machine learning and analytical tools into a live intrusion detection and prevention solution. This approach is named the Experimental Cyber Attack Detection Framework (ECAD). The purpose of this framework is to facilitate research of on-the-fly security applications. By integrating offline results in real-time traffic analysis, we could determine the type of network access as a legitimate or attack pattern, and discard/drop the latter. The results are promising and show the benefits of such a tool in the early prevention stages of both known and unknown cyber-attack patterns.

Cybersecurity plays an increasing role in today’s digital space, and its methods must keep pace with the changes. Both public and private sector researchers have put efforts into strengthening the security of networks by proposing new approaches. This paper presents a method to solve a game theory model by defining the contents of the game payoff matrix and incorporating honeypots in the defense strategy. Using a probabilistic approach we propose the course-of-action Stackelberg game (CoASG), where every path of the graph leads to an undesirable state based on security issues found in every host. The reality of the system is represented by a cost function which helps us to define a payoff matrix and find the best possible combination of the strategies once the game is run. The results show the benefits of using this model in the early prevention stages for detecting cyberattack patterns.

This paper presents a comprehensive study on the implementation of a Security-Enhanced Software-Defined Networking (SEC-SDN) framework integrated with game theory modeling and Snort, an open-source intrusion detection system, to strengthen cloud environments against Distributed Denial-of-Service (DDoS) attacks. As cloud computing continues to dominate the technological landscape, ensuring robust security for these highly scalable yet vulnerable systems is critical. The proposed SEC-SDN framework incorporates game theory to model dynamic interactions between attackers and defenders, enabling proactive and adaptive decision making for threat mitigation. By integrating Snort with SEC-SDN, the framework gains enhanced monitoring capabilities, facilitating real-time detection of malicious traffic patterns and the immediate enforcement of security policies to neutralize threats. This synergistic approach not only limits the impact of DDoS attacks but also preserves the availability and integrity of cloud services. A series of simulated DDoS attack scenarios were conducted on a cloud-based network to evaluate the effectiveness of the proposed framework. The results demonstrate significant improvements in threat detection, response times, and overall resilience of cloud environments against cyber threats. The findings highlight the potential of combining SEC-SDN, game theory modeling, and intrusion detection systems such as Snort to establish a scalable, adaptive, and robust security framework. This study contributes to the advancement of cloud security by proposing an innovative solution that dynamically evolves to address the increasingly sophisticated tactics of cyber attackers, ensuring the reliability and continuity of cloud services.

In our current paper we are introducing a new method to enhance semantic image segmentation accuracy of a U-Net neural network model by integrating it with a mathematical model based on reaction-diffusion equations. The methods currently used for semantic image segmentation, including U-Net neural networks, are processing images as blocks of pixels in which the boundaries, the colors and patterns are all mixed together as inputs to the transformations that take place inside the layers of the convolutional neural networks. In our method we are modifying the architecture of a U-Net network and introduce a new data input feed in parallel to the image feed that needs to be segmented. The new input feed is mathematically extracted from the input image and contains the edges (shape) information of the image to be processed. The new input feed it's used during the U-Net decoding phase in order to help shape more precisely the up-scaled output edges, thus leading to improved accuracy performance of the network. Introducing the parallel feed shows an improvement of accuracy metrics up to 4% (if compared to the U-Net model) and has a limited impact on computational resources consumed at training, because we are only adding a small number of new parameters to be calculated.

This paper describes the work of an objectual framework designed to be used in the parallelization of a set of related algorithms. As a concrete application a parallel Ant Colony Optimization algorithm (ACO) for the Travelling Salesman Problem (TSP) is presented. The idea behind the system we are describing is to have a re-usable framework for running several sequential algorithms in a parallel environment. The algorithms that the framework can be used with have several things in common: they have to run in cycles and the work should be possible to be split between several "processing units". The parallel framework uses the message-passing communication paradigm and is organized as a master-slave system. The ACO for TSP implemented by means of the parallel framework proves to have good performances: approximatively linear speedup and low communication cost.