arXiv subjects

Georgios Feretzakis

Publications and source records attributed to Georgios Feretzakis.

10 recordsLinked to original sources

arXiv2026

Signed but Steerable: Measuring and Closing the Text-Vector Integrity Gap in Offline Retrieval-Augmented Generation

Offline retrieval-augmented generation (RAG) on edge devices is updated through signed knowledge packs that carry text and precomputed embedding vectors. A signature proves who published a pack, not that its vectors encode its text. We study this text-vector integrity gap. A verifier that recomputes vectors and accepts them within a cosine tolerance concedes a predictable share of queries to an attacker who controls the vectors: the hijack rate equals the distribution of angular retrieval margins at the tolerance (the margin law). Margins are small in five corpora and two embedding models, so an idealised tolerance at the level of floating-point noise leaves up to 5.5% of queries exposed (at most 1.2% in any corpus once exact duplicate documents are excluded), and a threshold calibrated from our CPU INT8-FP32 comparison of a small embedding model (at the empirical 0.1% quantile of honest cosines) leaves 98-100% exposed; on a Raspberry Pi 5 running a local language model, such an attack lowers correct answers from 0.75 to 0.39. Integrity audits of compressed vector stores fare no better: on a FAISS product-quantisation index over 57,638 documents, tampered vectors that all pass the audit put the attacker's document first for 25% of targets. Verifying quantisation codes instead, with ambiguity decided by the verifier, a calibrated guard band and canonicalisation, accepted every honest vector in our samples on three platforms and left the publisher no choice. We also show that common floating-point and dynamic-INT8 inference pipelines are not bit-reproducible across batch sizes or architectures, and report pre-registered negative results on abstention, context pruning and semantic screening. We release code and data.

cs.CR↗

arXiv2026

Trust Mechanisms for Offline Generative AI at the Edge: A Connectivity-Regime Framework and an Empirical Study on a Raspberry Pi 5

Generative AI and large language models (LLMs) are moving into the Internet of Things, yet most deployments assume cloud connectivity, and many IoT settings have intermittent, restricted, or no connectivity. We propose a connectivity-regime framework (C0-C4), a taxonomy of inference-time trust mechanisms, and signed, versioned knowledge packs as the update channel for disconnected regimes, and evaluate them on a Raspberry Pi 5 running fully local retrieval-augmented generation under network isolation, with three Qwen2.5 model sizes, three context sizes, and 300 SQuAD 2.0 questions. An empirically calibrated retrieval-score gate detects knowledge gaps almost perfectly (AUROC 0.97) but in-topic unanswerable questions poorly (0.68). A deterministic output validator converts malformed outputs into explicit abstentions with no observed loss of correct answers, while the NLI-based verifier we tested adds little; for the 7B model, 96% of substantive answers to unanswerable-labelled questions match normalized spans of the retrieved passages. A 3B model achieves the same gold-answer containment rate as a 7B model at less than half the latency and estimated energy, although the 7B model far more often exceeds a token-F1 threshold of 0.5; five retrieved passages lower correctness from 75% to 52% compared with three. Verifying a signed pack takes about 4.5 ms versus about 29 min for embedding its passages on the device; signature and hash checks alone accept six of nine tested update attacks that layered verification rejects. Yet a validly signed poisoned pack passes every pack check: on 30 previously correct questions, correct answers drop to zero and 14 return the injected answer. None of the mechanisms we evaluated establishes the truth of the evidence. We release code and data.

cs.CR↗

arXiv2026

The Price of Peeking: Anytime-Valid Leakage Detection on ML-KEM EM Traces

Side-channel evaluators routinely inspect leakage tests while acquisition is still running, and extend or stop the campaign based on what they see. Fixed-horizon screening such as the Welch $t$-test with threshold $|t|>4.5$ gives no error guarantee for this monitored decision rule. We study anytime-valid leakage detection based on testing by betting: SKIT-type swap-pair e-processes whose false-alarm probability is controlled uniformly over time under an explicit conditional symmetry null. In matched comparisons that share the frozen witness, rows and payoff, first-crossing detection needed 1.68-2.00$\times$ the traces of a fixed-horizon randomization test at 80% detection on synthetic streams, and 1.68-2.38$\times$ on degraded recordings from an open ML-KEM electromagnetic dataset with the primary Ridge witness at $α=0.05$. With the same primary witness and level, on undegraded reference and pqm4 recordings the monitored procedure stopped early: its median stopping point was 62-72 and 146-316 evaluation traces, i.e. 2-8% of a conservative 4096-trace budget. Under exact designed nulls on the recorded backgrounds, repeated-look $|t|>4.5$ screening over all 13000-20000 samples raised a false alarm in 2.7-12.9% of replicates, against 0.0-4.7% for terminal-only screening and no rejection by a sample-wise e-Bonferroni process, which in a prespecified follow-up detected natural-label associations in 4 of 4 backgrounds after 840-3288 traces. All recordings come from one device, and natural-label results are descriptive; we state the assumptions each claim requires.

cs.CR↗

arXiv2025

Validating Emergency Department Admission Predictions Based on Local Data Through MIMIC-IV

The effective management of Emergency Department (ED) overcrowding is essential for improving patient outcomes and optimizing healthcare resource allocation. This study validates hospital admission prediction models initially developed using a small local dataset from a Greek hospital by leveraging the comprehensive MIMIC-IV dataset. After preprocessing the MIMIC-IV data, five algorithms were evaluated: Linear Discriminant Analysis (LDA), K-Nearest Neighbors (KNN), Random Forest (RF), Recursive Partitioning and Regression Trees (RPART), and Support Vector Machines (SVM Radial). Among these, RF demonstrated superior performance, achieving an Area Under the Receiver Operating Characteristic Curve (AUC-ROC) of 0.9999, sensitivity of 0.9997, and specificity of 0.9999 when applied to the MIMIC-IV data. These findings highlight the robustness of RF in handling complex datasets for admission prediction, establish MIMIC-IV as a valuable benchmark for validating models based on smaller local datasets, and provide actionable insights for improving ED management strategies.

cs.LG↗

arXiv2025

Semantic Synergy: Unlocking Policy Insights and Learning Pathways Through Advanced Skill Mapping

This research introduces a comprehensive system based on state-of-the-art natural language processing, semantic embedding, and efficient search techniques for retrieving similarities and thus generating actionable insights from raw textual information. The system automatically extracts and aggregates normalized competencies from multiple documents (such as policy files and curricula vitae) and creates strong relationships between recognized competencies, occupation profiles, and related learning courses. To validate its performance, we conducted a multi-tier evaluation that included both explicit and implicit skill references in synthetic and real-world documents. The results showed near-human-level accuracy, with F1 scores exceeding 0.95 for explicit skill detection and above 0.93 for implicit mentions. The system thereby establishes a sound foundation for supporting in-depth collaboration across the AE4RIA network. The methodology involves a multi-stage pipeline based on extensive preprocessing and data cleaning, semantic embedding and segmentation via SentenceTransformer, and skill extraction using a FAISS-based search method. The extracted skills are associated with occupation frameworks (as formulated in the ESCO ontology) and with learning paths offered through the Sustainable Development Goals Academy. Moreover, interactive visualization software, implemented with Dash and Plotly, presents graphs and tables for real-time exploration and informed decision-making by those involved in policymaking, training and learning supply, career transitions, and recruitment. Overall, this system, backed by rigorous validation, offers promising prospects for improved policymaking, human resource development, and lifelong learning by providing structured and actionable insights from raw, complex textual information.

cs.AI↗

arXiv2024

Trustworthy AI: Securing Sensitive Data in Large Language Models

Large Language Models (LLMs) have transformed natural language processing (NLP) by enabling robust text generation and understanding. However, their deployment in sensitive domains like healthcare, finance, and legal services raises critical concerns about privacy and data security. This paper proposes a comprehensive framework for embedding trust mechanisms into LLMs to dynamically control the disclosure of sensitive information. The framework integrates three core components: User Trust Profiling, Information Sensitivity Detection, and Adaptive Output Control. By leveraging techniques such as Role-Based Access Control (RBAC), Attribute-Based Access Control (ABAC), Named Entity Recognition (NER), contextual analysis, and privacy-preserving methods like differential privacy, the system ensures that sensitive information is disclosed appropriately based on the user's trust level. By focusing on balancing data utility and privacy, the proposed solution offers a novel approach to securely deploying LLMs in high-risk environments. Future work will focus on testing this framework across various domains to evaluate its effectiveness in managing sensitive data while maintaining system efficiency.

cs.AI↗

arXiv2021

Using machine learning techniques to predict hospital admission at the emergency department

Introduction: One of the most important tasks in the Emergency Department (ED) is to promptly identify the patients who will benefit from hospital admission. Machine Learning (ML) techniques show promise as diagnostic aids in healthcare. Material and methods: We investigated the following features seeking to investigate their performance in predicting hospital admission: serum levels of Urea, Creatinine, Lactate Dehydrogenase, Creatine Kinase, C-Reactive Protein, Complete Blood Count with differential, Activated Partial Thromboplastin Time, D Dimer, International Normalized Ratio, age, gender, triage disposition to ED unit and ambulance utilization. A total of 3,204 ED visits were analyzed. Results: The proposed algorithms generated models which demonstrated acceptable performance in predicting hospital admission of ED patients. The range of F-measure and ROC Area values of all eight evaluated algorithms were [0.679-0.708] and [0.734-0.774], respectively. Discussion: The main advantages of this tool include easy access, availability, yes/no result, and low cost. The clinical implications of our approach might facilitate a shift from traditional clinical decision-making to a more sophisticated model. Conclusion: Developing robust prognostic models with the utilization of common biomarkers is a project that might shape the future of emergency medicine. Our findings warrant confirmation with implementation in pragmatic ED trials.

cs.LG↗

arXiv2021

Pulmonary embolism identification in computerized tomography pulmonary angiography scans with deep learning technologies in COVID-19 patients

The main objective of this work is to utilize state-of-the-art deep learning approaches for the identification of pulmonary embolism in CTPA-Scans for COVID-19 patients, provide an initial assessment of their performance and, ultimately, provide a fast-track prototype solution (system). We adopted and assessed some of the most popular convolutional neural network architectures through transfer learning approaches, to strive to combine good model accuracy with fast training. Additionally, we exploited one of the most popular one-stage object detection models for the localization (through object detection) of the pulmonary embolism regions-of-interests. The models of both approaches are trained on an original CTPA-Scan dataset, where we annotated of 673 CTPA-Scan images with 1,465 bounding boxes in total, highlighting pulmonary embolism regions-of-interests. We provide a brief assessment of some state-of-the-art image classification models by achieving validation accuracies of 91% in pulmonary embolism classification. Additionally, we achieved a precision of about 68% on average in the object detection model for the pulmonary embolism localization under 50% IoU threshold. For both approaches, we provide the entire training pipelines for future studies (step by step processes through source code). In this study, we present some of the most accurate and fast deep learning models for pulmonary embolism identification in CTPA-Scans images, through classification and localization (object detection) approaches for patients infected by COVID-19. We provide a fast-track solution (system) for the research community of the area, which combines both classification and object detection models for improving the precision of identifying pulmonary embolisms.

eess.IV↗

arXiv2017

On Using Linear Diophantine Equations to Tune the extent of Look Ahead while Hiding Decision Tree Rules

This paper focuses on preserving the privacy of sensitive pat-terns when inducing decision trees. We adopt a record aug-mentation approach for hiding sensitive classification rules in binary datasets. Such a hiding methodology is preferred over other heuristic solutions like output perturbation or crypto-graphic techniques - which restrict the usability of the data - since the raw data itself is readily available for public use. In this paper, we propose a look ahead approach using linear Diophantine equations in order to add the appropriate number of instances while minimally disturbing the initial entropy of the nodes.

cs.AI↗

arXiv2017

Data set operations to hide decision tree rules

This paper focuses on preserving the privacy of sensitive patterns when inducing decision trees. We adopt a record augmentation approach for hiding sensitive classification rules in binary datasets. Such a hiding methodology is preferred over other heuristic solutions like output perturbation or cryptographic techniques - which restrict the usability of the data - since the raw data itself is readily available for public use. We show some key lemmas which are related to the hiding process and we also demonstrate the methodology with an example and an indicative experiment using a prototype hiding tool.

cs.AI↗