Theo Dimitrakos

Koffi Ismael Ouattara, Ioannis Krontiris, Theo Dimitrakos et al.

Trustworthiness in neural networks is crucial for their deployment in critical applications, where reliability, confidence, and uncertainty play pivotal roles in decision-making. Traditional performance metrics such as accuracy and precision fail to capture these aspects, particularly in cases where models exhibit overconfidence. To address these limitations, this paper introduces a novel framework for quantifying the trustworthiness of neural networks by incorporating subjective logic into the evaluation of Expected Calibration Error (ECE). This method provides a comprehensive measure of trust, disbelief, and uncertainty by clustering predicted probabilities and fusing opinions using appropriate fusion operators. We demonstrate the effectiveness of this approach through experiments on MNIST and CIFAR-10 datasets, where post-calibration results indicate improved trustworthiness. The proposed framework offers a more interpretable and nuanced assessment of AI models, with potential applications in sensitive domains such as healthcare and autonomous systems.

Koffi Ismael Ouattara, Ioannis Krontiris, Theo Dimitrakos et al.

As AI systems increasingly rely on training data, assessing dataset trustworthiness has become critical, particularly for properties like fairness or bias that emerge at the dataset level. Prior work has used Subjective Logic to assess trustworthiness of individual data, but not to evaluate trustworthiness properties that emerge only at the level of the dataset as a whole. This paper introduces the first formal framework for assessing the trustworthiness of AI training datasets, enabling uncertainty-aware evaluations of global properties such as bias. Built on Subjective Logic, our approach supports trust propositions and quantifies uncertainty in scenarios where evidence is incomplete, distributed, and/or conflicting. We instantiate this framework on the trustworthiness property of bias, and we experimentally evaluate it based on a traffic sign recognition dataset. The results demonstrate that our method captures class imbalance and remains interpretable and robust in both centralized and federated contexts.

Koffi Ismael Ouattara, Ioannis Krontiris, Theo Dimitrakos et al.

Trustworthiness has become a key requirement for the deployment of artificial intelligence systems in safety-critical applications. Conventional evaluation metrics, such as accuracy and precision, fail to appropriately capture uncertainty or the reliability of model predictions, particularly under adversarial or degraded conditions. This paper introduces the Parallel Trust Assessment System (PaTAS), a framework for modeling and propagating trust in neural networks using Subjective Logic (SL). PaTAS operates in parallel with standard neural computation through Trust Nodes and Trust Functions that propagate input, parameter, and activation trust across the network. The framework defines a Parameter Trust Update mechanism to refine parameter reliability during training and an Inference-Path Trust Assessment (IPTA) method to compute instance-specific trust at inference. Experiments on real-world and adversarial datasets demonstrate that PaTAS produces interpretable, symmetric, and convergent trust estimates that complement accuracy and expose reliability gaps in poisoned, biased, or uncertain data scenarios. The results show that PaTAS effectively distinguishes between benign and adversarial inputs and identifies cases where model confidence diverges from actual reliability. By enabling transparent and quantifiable trust reasoning within neural architectures, PaTAS provides a foundation for evaluating model reliability across the AI lifecycle.

Pierre de Leusse, Theo Dimitrakos

Business requirements for rapid operational efficiency, customer responsiveness as well as rapid adaptability are actively driving the need for ever increasing communication and integration apabilities of software assets. In this context, security, although acknowledged as being a necessity, is often perceived as a hindrance. Indeed, dynamic environments require flexible and understandable security that can be customized, adapted and reconfigured dynamically to face changing requirements. In this paper, the authors propose SOA based security governance middleware that handles security requirements on behalf of a resource exposed through it. The middleware aims at providing different security settings through the use of managed compositions of security services called profiles. The main added value of this work compared to existing handlers or centralized approaches lies in its enhanced flexibility and transparency.

Pierre de Leusse, Panos Periorellis, Theo Dimitrakos et al.

The Internet of Things and Services is a rapidly growing concept that illustrates that the ever increasing amount of physical items of our daily life which become addressable through a network could be made more easily manageable and usable through the use of Services. This surge of exposed resources along with the level of privacy and value of the information they hold, together with the increase of their usage make for an augmentation in the number of the security threats and violation attempts that existing security systems do not appear robust enough to address. In this paper, the authors underline this increase in risk and identify the requirements for resources to be more resilient in this type of environment while keeping an important level of flexibility. In addition, the authors propose an architectural model of Self Managed Security Cell, which leverages on current knowledge in large scale security systems, information management and autonomous systems.