Skip to main content

MEJORANDO LA DETECCION DE CIBER AMENAZAS USANDO MODELOS DE LENGUAJE DE GRAN TAMAÑO PARA PROTOCOLOS DE RED

PID2023-150794OB-I00

•
Funding agency name Agencia Estatal de Investigación
Funding agency acronym AEI
Program Programa Estatal para Impulsar la Investigación Científico-Técnica y su Transferencia
Subprogram Subprograma Estatal de Generación de Conocimiento
Call Proyectos de I+D+i (Generación de Conocimiento y Retos Investigación)
Call year 2023
State Plan 2021-2023 Plan Estatal de Investigación Científica y Técnica y de Innovación
Beneficiary centre UNIVERSIDADE DA CORUÑA

Resultados relacionados

Resultados totales (Incluyendo duplicados): 9
Encontrada(s) 1 página(s)

RUC. Repositorio da Universidade da Coruña
Publication Artículo científico (JournalArticle)

Building a Supervised–Unsupervised Ensemble for Intrusion Detection Systems Using Model Calibration and Decision Fusion

  • Lagares Vázquez, Ángel
  • Garabato, D.
  • Carneiro, Víctor
  • Fernández, Diego
  • Cacheda, Fidel
  • Nóvoa, Francisco
Financiado para publicación en acceso aberto: Universidade da Coruña/CISUG, [Abstract] Machine learning, particularly deep learning, has become a prominent approach to network intrusion detection because it can learn patterns from network traffic and adapt to evolving threats. However, most existing Intrusion Detection Systems (IDS) approaches address only one side of the problem: they either classify known attacks from labeled data or detect previously unseen attacks through anomaly detection. In operational network environments, both capabilities are required within a single adaptive system. To address this limitation, we propose a hybrid IDS that combines supervised and unsupervised models in a calibration-aware ensemble designed to detect both known attacks and attack classes absent from the training data. The main contribution of this work is a calibration and decision fusion pipeline that aligns the outputs of heterogeneous models and integrates them into a unified detection decision. Experiments conducted using two benchmark datasets, CIC-IDS2017 and BoT-IoT, across different experimental settings, show that the proposed ensemble is capable of both detecting previously unseen attacks and performing multiclass classification of known attacks, while maintaining comparable multiclass performance despite the increased task complexity. It also improves binary detection performance, indicating that the hybrid design successfully exploits the complementary strengths of supervised classification and unsupervised anomaly detection., This work was carried out at CITIC, within the framework of the project PID2023-150794OB-I00, funded by the Ministry of Science, Innovation and Universities (MICIU) and the State Research Agency (AEI) /10.13039/501100011033, and co-funded by the European Regional Development Fund (ERDF), European Union. We also acknowledge support from the Xunta de Galicia and the European Union (FEDER Galicia 2021–2027 Program) through grants ED431B 2024/21, ED431B 2024/02 and CITIC ED431G 2023/01. Partially funding was also received through ERDF EU Interreg VI-A Spain-Portugal (POCTEP) 2021–2027 Programme, project “Quantum IBER_IA: Impulso estratégico de las capacidades en tecnologías cuánticas e inteligencia artificial en el espacio ibérico transfronterizo”. Funding for open access charge: Universidade da Coruña/CISUG., Xunta de Galicia; ED431B 2024/21, Xunta de Galicia; ED431B 2024/02, Xunta de Galicia; ED431G 2023/01
RUC. Repositorio da Universidade da Coruña
Publication Artículo científico (JournalArticle)

SDN-CF: Traffic classification in SDN ONOS controller using machine learning models

  • Carneiro, Víctor
  • Álvarez, M. A.
  • Cacheda, Fidel
Permanent link to code/repository used for this code version: https://github.com/ElsevierSoftwareX/SOFTX-D-25-00379, [Abstract]: SDN-CF (Software-Defined Network - Classification Framework) is a modular Java-based application built on the Northbound API of the ONOS Software-Defined Network (SDN) controller for network traffic analysis using machine learning techniques. While it employs the Random Forest algorithm by default, its open design allows the integration of alternative classifiers. SDN-CF enables the dynamic blocking of unwanted connections and generates an annotated dataset of OpenFlow traffic, supporting reproducible research in anomaly detection. Designed for academic and experimental use in virtualized environments, the tool fosters the evaluation and development of novel detection approaches in SDN contexts., This work was carried out at CITIC, within the framework of the project PID2023-150794OB-I00, funded by the Ministry of Science, Innovation and Universities (MICIU) and the State Research Agency (AEI) /10.13039/501100011033, and co-funded by the European Regional Development Fund (ERDF), European Union. CITIC, accredited as a center of excellence within the Galician University System and a member of the CIGUS Network, also receives support from the Department of Education, Science, Universities, and Vocational Training of the Xunta de Galicia, co-financed by the EU through the ERDF Galicia 2021–2027 programme (Ref. ED431G 2023/01)., Xunta de Galicia; ED431G 2023/01
RUC. Repositorio da Universidade da Coruña
Publication Artículo científico (JournalArticle)

Automated legal analysis of rental contract clauses using large language models

  • Carneiro, Víctor
  • Grille-Zallas, Adrián
  • Lage-Etchart, Daniel
Permanent link to code/repository used for this code version: https://github.com/ElsevierSoftwareX/SOFTX-D-25-00431 Permanent link to Reproducible Capsule: https://github.com/cyberLLM-Net/SLC-GPT/tree/main/library, [Abstract]: This work presents a software tool for the automated legal analysis of rental contract clauses using large language models (LLMs). The system extracts clause-level segments from .docx files and generates legal evaluations and rewritten alternatives using multiple LLMs (GPT-4, GPT-4o, GPT-3.5, DeepSeek). Each clause is assessed for legality, abusiveness, vagueness, risk, and legal gaps according to Spanish housing law. A unified JSON structure stores the outputs for reproducibility and benchmarking. The tool enables comparative analysis across models and provides a foundation for future research on AI-assisted legal drafting and clause evaluation., This work was carried out at CITIC, within the framework of the project PID2023-150794OB-I00, funded by the Ministry of Science, Innovation and Universities (MICIU), and co-funded by the European Regional Development Fund (ERDF), European Union. CITIC, accredited as a center of excellence within the Galician University System and a member of the CIGUS Network, also receives support from the Department of Education, Science, Universities, and Vocational Training of the Xunta de Galicia, co-financed by the EU through the ERDF Galicia 2021–2027 programme (Ref. ED431G 2023/01)., Xunta de Galicia; ED431G 2023/01
RUC. Repositorio da Universidade da Coruña
Publication Artículo científico (JournalArticle)

UDCDGA: A Reproducible Software Framework for Constructing Large-scale DGA Detection Datasets

  • Carneiro, Víctor
  • Álvarez, M. A.
  • Padín Torrente, Héctor
  • Bustos-Jiménez, Javier
Code/repository: https://github.com/cyberLLM-Net/DGA-xAI/tree/v1.0.1, [Abstract] This work introduces UDCDGA_Generator, a modular Python framework for the reproducible construction of large-scale datasets for Domain Generation Algorithm (DGA) detection research. The software integrates benignsource aggregation, execution of heterogeneous DGA implementations, deterministic sampling and balancing, global deduplication and cross-class overlap removal, feature extraction, integrity validation, and dataset export within a configurable workflow. DGA-generated domains are produced through an adaptive generation pipeline that supports configurable seed and date strategies, monitors DGA implementation execution, and enforces global uniqueness. The dataset construction module applies a default registry of 54 numerical features spanning lexical, structural, statistical, and n-gram-based representations, while supporting experiment-specific feature config urations and generating metadata and integrity artifacts for traceability. The capabilities of the framework are illustrated through the construction of UDCDGA, a balanced dataset containing 8,450,556 unique domain names, equally distributed between benign-source and DGA-generated domains, and exported in Apache Parquet format together with metadata describing its schema, integrity, and construction process. By decoupling dataset con struction from fixed domain corpora, UDCDGA_Generator provides configurable and extensible infrastructure for constructing reproducible datasets for DGA detection research., This work was carried out at CITIC, within the framework of the project PID2023-150794OB-I00, funded by the Ministry of Science, Innovation and Universities (MICIU) and the State Research Agency (AEI) /10.13039/501100011033, and co-funded by the European Regional Development Fund (ERDF), European Union. CITIC, accredited as a center of excellence within the Galician University System and a member of the CIGUS Network, also receives support from the Department of Education, Science, Universities, and Vocational Training of the Xunta de Galicia, co-financed by the EU through the European Regional Development Fund (ERDF) – Galicia 2021–2027 programme (Ref. ED431G 2023/01)., Xunta de Galicia; ED431G 2023/01, https://github.com/cyberLLM-Net/DGA-xAI/tree/v1.0.1
RUC. Repositorio da Universidade da Coruña
Publication Artículo científico (JournalArticle)

A Hybrid Deep Learning-Based Architecture for Network Traffic Anomaly Detection via EFMS-Enhanced KMeans Clustering and CNN-GRU Models

  • Quirumbay Yagua, Daniel
  • Fernández, Diego
  • Nóvoa, Francisco
[Abstract]: Early detection of network traffic anomalies is critical for cybersecurity, as a single compromised host can cause data breaches, reputational damage, and operational disruptions. However, traditional systems based on signatures and static rules are often ineffective against sophisticated and evolving threats. This study proposes a hybrid deep learning architecture for proactive anomaly detection in local and metropolitan networks. The dataset underwent an extensive process of cleaning, transformation, and feature selection, including normalization of numerical fields, encoding of ordinal variables, and derivation of behavioral metrics. The EFMS-KMeans algorithm was applied to pre-label traffic as normal or anomalous by estimating dense centers and computing centroid distances, enabling the training of a sequential CNN-GRU network, where the CNN captures spatial patterns and the GRU models temporal dependencies. To address class imbalance, the SMOTE technique was integrated, and the loss function was adjusted to improve training stability. Experimental results show a substantial improvement in accuracy and generalization compared to conventional approaches, validating the effectiveness of the proposed method for detecting anomalous traffic in dynamic and complex network environments., This work was supported in part by the Ministry of Science, Innovation and Universities (Spanish National Authority for Scientific Research and Innovation) and FEDER Funds of the European Union under Project PID2023-150794OB-I00. We also acknowledge support from the Xunta de Galicia and the European Union (FEDER Galicia 2021-2027 Program) under Grants ED431B 2024/02, and ED431G 2023/01., Xunta de Galicia; ED431B 2024/02, Xunta de Galicia; ED431G 2023/01
RUC. Repositorio da Universidade da Coruña
Publication Artículo científico (JournalArticle)

A Data-Centric Network Traffic Dataset for Anomaly Detection: Construction, Reproducible Pipeline, and Technical Validation

  • Quirumbay Yagua, Daniel
  • Fernández, Diego
  • Nóvoa, Francisco
  • Garabato, D.
[Abstract]: The effectiveness of machine learning and deep learning methods for network anomaly detection depends strongly on the quality and representativeness of the datasets used for training and evaluation. Despite recent advances, many publicly available benchmarks rely on synthetic traffic, outdated attack scenarios, or limited representation of encrypted communications. This work presents a network traffic dataset derived from operational firewall logs collected in a heterogeneous institutional environment dominated by HTTPS/TLS traffic. A structured data-centric pipeline was implemented, including preprocessing, behavioral feature engineering, unsupervised pseudo-labeling through the EFMS–KMeans algorithm, class balancing using SMOTE, and the generation of model-oriented sequential representations for deep learning analysis. The resulting dataset contains large-scale flow-level records describing volumetric, behavioral, and temporal traffic characteristics while preserving privacy through anonymization procedures. Technical validation was conducted using statistical analysis, entropy-based measurements, clustering quality metrics, and dimensionality reduction techniques, confirming data consistency, structural diversity, and class separability. The dataset is publicly available through the Mendeley Data repository together with metadata and documentation supporting anomaly detection research, encrypted traffic analysis, and the evaluation of machine learning and deep learning approaches in realistic cybersecurity environments., This work was carried out at CITIC within the framework of project PID2023-150794OB-I00, funded by the Ministry of Science, Innovation and Universities (MICIU), the State Research Agency (AEI) (Grant No. 10.13039/501100011033), and co-funded by the European Regional Development Fund (ERDF), European Union. The authors also acknowledge support from the Xunta de Galicia and the European Union (FEDER Galicia 2021–2027 Programme) through grants ED431B 2024/21, ED431B 2024/02, and CITIC ED431G 2023/01. Partial funding was also received through the ERDF EU Interreg VI-A Spain–Portugal (POCTEP) 2021–2027 Programme under the project “Quantum IBER_IA: Impulso estratégico de las capacidades en tecnologías cuánticas e inteligencia artificial en el espacio ibérico transfronterizo”., Xunta de Galicia; ED431B 2024/21, Xunta de Galicia; ED431B 2024/02, Xunta de Galicia; ED431G 2023/01, http://dx.doi.org/10.17632/g5dtm7349f
RUC. Repositorio da Universidade da Coruña
Publication Artículo científico (JournalArticle)

Early Detection of Infectious Diseases Using Wearable Sensor Data through Personalized Baseline Deviation Modelling

  • Cacheda, Fidel
  • Carneiro, Víctor
  • Álvarez, M. A.
  • López-Vizcaíno, Manuel F.
The data used in this research is publicly available at: https://www.synapse.org/Synapse:syn22891469/wiki/, [Abstract]: Objective. Early detection of infectious diseases using physiological and behavioural signals from wearable devices could enable earlier intervention and help reduce disease transmission. This study proposes and evaluates a framework specifically designed for sequential early-warning aimed at detecting infectious disease episodes prior to clinical diagnosis. Methods. In this retrospective observational cohort study, we used the Evidation 2019–2020 Fitbit dataset to formulate infection detection as an episode-level prediction problem. We designed augmented features that capture short-term deviations from each individual’s healthy baseline over sliding windows and evaluated several machine learning models. The framework incorporates probability calibration, validation-based threshold selection, and an optional persistent two-day alarm rule to reduce false-positive alerts. Results. Experiments on data from 435 participants showed that personalized deviation-based features substantially outperformed basic wearable features. The best-performing configuration achieved an episode-level F1-score of 0.784 with an Average Earliness (AE) of 0.555, corresponding to approximately 5.5 days before diagnosis. Additionally, combining step and sleep features provided a favourable trade-off between timeliness (AE = 0.69) and accuracy (F1 = 0.75). Conclusions. Personalized baseline-deviation features, combined with calibrated probabilities and persistent alarm mechanisms, enable accurate and timely early detection of infectious disease episodes from wearable data. The results indicate that recent individualized baselines and step-derived signals provide the most robust predictive information, supporting the potential of wearable devices as scalable tools for population-level infectious disease surveillance., The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was carried out at CITIC, within the framework of the project PID2023-150794OB-I00. funded by the Ministry of Science, Innovation and Universities (MICIU) and the State Research Agency (AEI) /10.13039/501100011033, and co-funded by the European Regional Development Fund (ERDF), European Union. CITIC, accredited as a center of excellence within the Galician University System and a member of the CIGUS Network, also receives support from the Department of Education, Science, Universities, and Vocational Training of the Xunta de Galicia, co-financed by the EU through the ERDF Galicia 2021–2027 programme (Ref. ED431G 2023/01)., Xunta de Galicia; ED431G 2023/01
RUC. Repositorio da Universidade da Coruña
Publication Artículo científico (JournalArticle)

Touch-based Interaction Dataset for user Behavioral Analysis in Mobile Devices

  • Garabato, D.
  • Casado Diez, Mario
  • Dafonte, Carlos
  • López-Vizcaíno, Manuel F.
  • Álvarez, M. A.
  • Nóvoa, Francisco
Repository name: AITouch – Data (Mendeley Data), 10.17632/9v7bxv3dcc.1 Direct URL to data: https://data.mendeley.com/datasets/9v7bxv3dcc/1, [Abstract]: As smartphones and tablets are increasingly integrated into numerous aspects of everyday life, security on mobile devices is becoming important. Although facial recognition or fingerprint scanning are commonly employed to verify user identity, they are not always suitable for on-demand validation during specific moments of interaction. This limitation has motivated the search for alternative solutions, particularly those based on behavioral biometrics, as they enable the capture and analysis of unique interaction patterns without disrupting the user experience. This paper describes a dataset comprising touch-based interactions collected from 37 distinct users within a controlled ad-hoc scenario designated for authentication purposes. The dataset includes a wide range of touch gestures, from single-touch events (e.g., tap, swipe or pan) to multi-touch interactions (e.g., pinch or rotate), which can be utilized to extract individual behavioral patterns from user-device interactions, thus supporting further research. In particular, the data acquisition process is thoroughly described, so that raw data can be appropriately understood, as well as those features that were extracted to carry out our own research. Apart from the data being available on a public repository, we also include some base code that can help other researchers to handle raw data and extract the aforementioned features, so that they can adapt or even extend them according to their own needs., This work has been developed in part thanks to the grant TED2021–130492B-C21 funded by MCIN/AEI/10.13039/501100011033 and by “ERDF A way of making Europe”. This work is funded in part by Xunta de Galicia and the European Union (European Regional Development Fund–Galicia 2021–2027 Program), under Grant ED431B 2024/02 and Grant ED431B 2024/21. This work was also supported by grant number PID2023–150794OB-I00, funded by the  MICIU/AEI/10.13039/501100011033, and by “ERDF A way of making Europe”. We also acknowledge support from CIGUS-CITIC, funded by Xunta de Galicia and the European Union (ERDF–Galicia 2021–2027 Program), through grant ED431G 023/01., Xunta de Galicia; ED431B 2024/02, Xunta de Galicia; ED431B 2024/21, Xunta de Galicia; ED431G 023/01
RUC. Repositorio da Universidade da Coruña
Publication Artículo científico (JournalArticle)

Toward Human-Centered Explainability: Natural Language Explanations for Anomaly Detection

  • Padín Torrente, Héctor
  • Carneiro, Víctor
  • Ortega-Fernández, Inés
The benchmark dataset analyzed in this study is the CSE–CIC-IDS2018 maintained by the Canadian Institute for Cybersecurity, University of New Brunswick, openly accessible at https://www.unb.ca/cic/datasets/ids-2018.html. The evaluation data used to assess the models and the various data generation strategies consist of the predictions produced by the autoencoder in combination with SHAP outputs, applied to the attack instances within the dataset. This evaluation data, as well as prompt details, are available at GitHub (https://github.com/Gradiant/Natural-Language-Explanations-4-AD)., [Abstract]: This paper proposes a human-centered explainable artificial intelligence pipeline for anomaly detection, designed to generate meaningful, context-aware explanations using local large language models. The proposed pipeline translates model outputs and SHAP-based feature attributions into natural language explanations for cybersecurity alerts generated by an autoencoder within an enterprise network. It incorporates a human-in-the-loop component to ground the explanations in validated expert knowledge, enhancing their interpretability and alignment with human decision-making processes. Using a rubric-driven LLM-as-a-Judge evaluation, we benchmark several large language models and show that as smaller models receive more contextual grounding through human-in-the-loop, their explanatory performance improves significantly, narrowing the gap with larger models while maintaining substantially lower computational demands. Our approach provides targeted, context-aware explanations designed to meet the cognitive and operational needs of security analysts, contributing to more ethical, trustworthy, and resource-efficient AI integration in critical cybersecurity environments., Open Access funding provided thanks to the CRUE-CSIC agreement with Springer Nature. This publication is part of the project SafeNet UEBA (CPP001/23-R006_SafeNet_UEBA), financed by “European Union NextGeneration-EU, https://next-generation-eu.europa.eu/, the Recovery Plan, Transformation and Resilience, https://planderecuperacion.gob.es/, through INCIBE, https://www.incibe.es/”. This work was carried out at CITIC, within the framework of the project PID2023-150794OB-I00, funded by the Ministry of Science, Innovation and Universities (MICIU) and the State Research Agency (AEI)/10.13039/501100011033, and co-funded by the European Regional Development Fund (ERDF), European Union. CITIC, accredited as a center of excellence within the Galician University System and a member of the CIGUS Network, also receives support from the Department of Education, Science, Universities, and Vocational Training of the Xunta de Galicia, co-financed by the EU through the ERDF Galicia 2021–2027 programme (Ref. ED431G 2023/01)., Xunta de Galicia; ED431G 2023/01, Instituto Nacional de Ciberseguridad (INCIBE); CPP001/23-R006_SafeNet_UEBA