Information Technologies Institute

Abstract As cyber-attacks grow in scale, sophistication, and impact, advanced training and evaluation platforms are essential for cybersecurity readiness. Cyber Ranges (CRs) provide controlled, high-fidelity simulation environments in which professionals can practice defensive strategies under realistic threat scenarios. In such environments, success depends not only on technical skill but also on robust Situational Awareness (SA), the cognitive capacity to perceive events, understand their meaning, and project future developments. Although SA is recognised as critical, existing CR evaluation methods typically focus on overall performance metrics (e.g., task completion rates, response accuracy, mission success). Such assessments often fail to capture the deeper, cognitive aspects of SA, limiting the CR’s ability to evaluate decision-making agility and strategic insight. In response, this work presents the SA Scoring System (SASS), a dynamic, multi-dimensional framework designed to quantify SA in live CR environments. SASS decomposes SA into core dimensions, perception, comprehension, projection, and resolution, then monitors trainee behaviour and scenario interactions in real-time. Each dimension is assessed via weighted metrics: timing of responses, sequence and number of reasoning steps, usage and effectiveness of tools and techniques, hints requested, and decision outcomes. These metrics are aggregated into interpretable scores that reflect both aspects, specific performance and overall SA proficiency. In this study, the SASS is validated in a series of cyber-defence exercises involving participants of differing expertise levels. The results demonstrate that SASS scores correlate strongly with expert evaluations and can discriminate between high and low SA performers with high fidelity. SASS enables instructors to pinpoint cognitive weak spots (e.g. delayed perception or faulty projection) and adapt training on the fly. Ultimately, integrating the SASS into CR platforms enriches the evaluation process, promoting not just technical competence but also the situational agility needed to counter modern cyber threats.

Abstract Despite being a well-studied problem, bias continues to affect modern Computer Vision (CV) datasets, models and systems, including Generative AI models that have been found to replicate and amplify harmful biases and stereotypes. In this work, we focus on framing bias : a type of bias that arises from the way images convey meaning. We introduce a semiotic perspective that treats image datasets as texts and argues that framing bias is not a statistical accident due to sampling noise or imbalance, but an inherent property of image datasets as meaning-making systems. We show that co-occurrence of visual elements is the primary mechanism through which image datasets frame concepts, and that not only do individual images contribute to dataset framing through co-occurrence, but dataset-level framing also shapes the interpretation of individual images. As a consequence, every image dataset embodies a point of view and cannot be fully unbiased . We illustrate these claims through a case study of the Visual Genome dataset, revealing a framing of human activities that over-represents leisure, systematically excludes everyday labour, and is characterised by a predominantly American/Western viewpoint. Finally, we reflect on the epistemological implications of analysing image datasets, highlighting the mediated and interpretive nature of knowledge produced through AI systems, and propose revisionism and source criticism as epistemological paradigms to address these issues.

Purpose The purpose of this study is to explore how different types of price discounts influence consumers’ purchase intentions within a simulated supermarket environment. By integrating eye tracking (ET) and electroencephalography (EEG) technologies, the research aims to capture real-time visual attention and neural responses, offering insights beyond traditional self-report methods. Specifically, the study examines whether shopping duration and age predict purchasing behaviour under various discount types, while also assessing the role of memorization and fixation patterns in shaping consumer decisions. Design/methodology/approach To simulate a real-life shopping experience, a leaflet identical to those used in supermarkets was created for the study. ET and EEG were employed to monitor participants’ gaze and emotional responses while interacting with the leaflet. The behaviour of 42 participants is analysed to assess the clarity of promotional messages and different types of discounts. Additionally, EEG data provided insights into unconscious preferences and cognitive processes, such as memorization and approach-withdrawal tendencies, during decision-making. Findings The analysis reveals that Fixation Duration is the strongest predictor of purchasing behaviour, showing a robust positive association between the time spent viewing a product and the likelihood of purchase. EEG results indicate that Memorization significantly predicts buying decisions, while Attention-Withdrawal does not. Age and shopping duration contribute meaningfully to explaining variations in the number of products purchased, though their influence depends on discount type. Specifically, shopping duration predicts purchases under the “50% off” discount, whereas age is more relevant for the “1 + 1 free” offer. Neither factor predicts purchases when the vague “ONLY” discount is applied. Overall, the findings underscore the importance of visual engagement and cognitive processing in shaping consumer choices, while highlighting that clear and substantial discounts outperform ambiguous ones. Practical implications By understanding the factors that influence consumer decision-making, businesses can tailor their marketing strategies to improve product selection and boost sales. The study’s findings can help retailers design more effective promotional materials and optimize the layout of their physical stores or online shopping platforms. This, in turn, can shorten decision-making time and ensure that high-priority products are more prominently featured. Originality/value The study combines advanced neuroscience (EEG) with ET technology to analyse consumer behaviour, which adds a novel dimension to traditional marketing research. By using a supermarket leaflet identical to actual ones, the study replicates real-world shopping scenarios, ensuring realistic insights into consumer decision-making. The study focuses on cognitive and behavioural processes, and it explores deeper, unconscious factors like memorization and approach-withdrawal tendencies, which are often overlooked in consumer behaviour studies.

Software agent technology has matured enough to produce intelligent agents, which can be used to control a large number of concurrent engineering tasks. Multi-agent systems are communities of agents that exchange information and data in the form of messages. The agents’ intelligence can range from rudimentary sensor monitoring and data reporting to more advanced forms of decision making and autonomous behavior. The behavior and intelligence of each agent in the community can be obtained by performing data mining on available application data and the respected knowledge domain. We have developed Agent Academy, a software platform for the design, creation, and deployment of multi-agent systems, which combines the power of knowledge discovery algorithms with the versatility of agents. Using this platform, we show how agents, equipped with a data-driven inference engine, can be dynamically and continuously trained. We also discuss a few prototype multi-agent systems developed with Agent Academy.

To excel at their domain, large language models are comprised of billions of parameters. Yet, this comes at the cost of huge memory requirements, restricting their applicability in resource-constrained environments. To address the problem of neural network (NN) compression, Singular Value Decomposition (SVD) has played a key role as a fundamental component for matrix compression through decomposition. To minimize compression error and to maximize the efficacy of the compressed model on the downstream tasks, previous works focused on low-rank approximation of the NN’s weight matrices either from the perspective of parameter importance or per-layer functional equivalence. While previous works studied the aforementioned perspectives in isolation, in this work, we are investigating the effectiveness of an approach that combines ideas from these two perspectives in a single objective. In parallel to this, an important aspect that affects the compression quality is the distribution of the compression rate across layers and NN parameters. Earlier works mostly considered distributing the compression rate uniformly across layers and network weights or relied on computationally expensive heuristic search. Contrary to them, in this work, we propose an enhanced and computationally efficient algorithm for dynamic compression rate allocation. Experimental results support the efficacy of the proposed approach, which performs on par or substantially better than the previous state-of-the-art, especially under high compression ratios.

Machine learning (ML) is becoming a key enabler in building energy management systems (BEMS), yet most existing reviews focus on simulations and fail to reflect the realities of real-world deployment. In response to this limitation, the present work aims to present a systematic review dedicated entirely to experimental, field-tested applications of ML in BEMS, covering systems such as Heating, Ventilation & Air-conditioning (HVAC), Renewable Energy Systems (RES), Energy Storage Systems (ESS), Ground Heat Pumps (GHP), Domestic Hot Water (DHW), Electric Vehicle Charging (EVCS), and Lighting Systems (LS). A total of 73 real-world deployments are analyzed, featuring techniques like Model Predictive Control (MPC), Artificial Neural Networks (ANNs), Reinforcement Learning (RL), Fuzzy Logic Control (FLC), metaheuristics, and hybrid approaches. In order to cover both methodological and practical aspects, and properly identify trends and potential challenges in the field, current review uses a unified framework: On the methodological side, it examines key-attributes such as algorithm design, agent architectures, data requirements, baselines, and performance metrics. From a practical standpoint, the study focuses on building typologies, deployment architectures, zones scalability, climate, location, and experimental duration. In this context, the current effort offers a holistic overview of the scientific landscape, outlining key trends and challenges in real-world machine learning applications for BEMS research. By focusing exclusively on real-world implementations, this study offers an evidence-based understanding of the strengths, limitations, and future potential of ML in building energy control—providing actionable insights for researchers, practitioners, and policymakers working toward smarter, grid-responsive buildings. Findings reveal a maturing field with clear trends: MPC remains the most deployment-ready, ANNs provide efficient forecasting capabilities, RL is gaining traction through safer offline–online learning strategies, FLC offers simplicity and interpretability, and hybrid methods show strong performance in multi-energy setups.

The use of Unmanned Aerial Vehicles (UAVs) in various domains is continuously increasing. The benefits offered by UAVs enable the facilitation of functions or execution of tasks, such as search and rescue, which would otherwise be challenging or financially restrictive. Furthermore, UAVs are important for the military domain being considered as efficient weapons. Therefore, many countries and retail companies are investing in the massive production of UAVs for market as well as government purposes. Nevertheless, most UAVs in all domains, lack even basic cybersecurity mechanisms, thus, posing cybersecurity threats that significantly affect cybersecurity aspects or result in physical damage and even loss of human life. Many cybersecurity solutions have been proposed in the literature, encompassing both reactive and proactive measures, with the latter being preferable. Added to this, threat modeling emerges as an effective proactive measure with minimal cost and complexity. The current work conducts extensive and comprehensive research concerning the domain of UAVs. Initially, the paper outlines the historical development of UAVs and the different taxonomies in which they are classified according to different characteristics (e.g., number of rotors). The next sections detail a comprehensive description of the Internet of Drones (IoD) environment and the various relevant cybersecurity issues. Subsequently, the paper examines a wide range of diverse threat modeling approaches found in the literature, while it categorises relevant papers based on whether they describe solutions that integrate threat modeling or propose novel approaches for threat modeling.