Publications

@article{redondo-gutierrez_detecting_2022,

title = {Detecting malware using text documents extracted from spam email through machine learning},

author = {Luis Ángel Redondo-Gutierrez and Francisco Jáñez-Martino and Eduardo Fidalgo and Enrique Alegre and Víctor González-Castro and Rocío Alaiz-Rodríguez},

url = {https://dl.acm.org/doi/abs/10.1145/3558100.3563854},

year  = {2022},

date = {2022-01-01},

journal = {Proceedings of the 22nd ACM Symposium on Document Engineering},

pages = {1–4},

abstract = {This work introduces the "Spam Email Malware Detection - 600" (SEMD-600) dataset for detecting malware in spam emails using text analysis. It compares two text representation techniques (Bag of Words and TF-IDF) combined with three classifiers (SVM, Naive Bayes, and Logistic Regression). The combination of TF-IDF and Logistic Regression achieved the best performance, with a macro F1 score of 0.763.},

keywords = {Malware Detection, NLP, Spam Email, Text classification},

pubstate = {published},

tppubtype = {article}

}

@article{castano_evaluation_2022,

title = {Evaluation of supervised learning models using TCP traffic for the detection of botnets},

author = {Felipe Castaño and Javier Velasco-Mata and Andrés Carofilis-Vasco and Eduardo Fidalgo and Laura Fernández-Robles and George Azzopardi},

url = {https://dialnet.unirioja.es/servlet/articulo?codigo=9206645},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {VII Jornadas Nacionales de Investigación en Ciberseguridad 2022},

abstract = {Este trabajo propone un conjunto de 23 variables para clasificar la severidad de incidentes de ciberseguridad asociados a direcciones IP. Se utilizan series temporales, listas de reputación y datos de geolocalización para construir un modelo multi-clase basado en aprendizaje automático. Además, se realiza un análisis estadístico para optimizar y validar la adecuación de las variables propuestas.},

keywords = {Aprendizaje Automático, Direcciones IP, Selección de Variables, Severidad},

pubstate = {published},

tppubtype = {article}

}

@article{castano_extracting_2022,

title = {Extracting Composition and Social Engineering Features to Measure Spam Address Credibility},

author = {Felipe Castaño and Francisco Jañez-Martino and Pablo Blanco-Medina and Alexandra Bonnici and Santiago González and Eduardo Fidalgo},

url = {https://dialnet.unirioja.es/servlet/articulo?codigo=9206644},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {VII Jornadas Nacionales de Investigación en Ciberseguridad 2022},

abstract = {Este trabajo propone un modelo para clasificar la severidad de incidentes de ciberseguridad mediante aprendizaje automático. Se construyó un conjunto de 23 variables que caracterizan la maliciosidad de una dirección IP, combinando datos de series temporales, listas de reputación y geolocalización. Se realizó un análisis estadístico para validar y optimizar estas variables, considerando su estabilidad en el tiempo y sensibilidad a hiperparámetros.},

keywords = {Aprendizaje Automático, Direcciones IP, Selección de Variables, Severidad},

pubstate = {published},

tppubtype = {article}

}

@article{del_rio_ferreras_semantic_2022,

title = {Semantic Attention Keypoint Filtering for Darknet Content Classification},

author = {Aitor Del Río Ferreras and Eduardo Fidalgo and Pablo Blanco-Medina and Deisy Chaves and Alexci Prieto-Castro and Enrique Alegre},

url = {https://dialnet.unirioja.es/servlet/articulo?codigo=9206652},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {VII Jornadas Nacionales de Investigación en Ciberseguridad 2022},

abstract = {Investigación en Ciberseguridad Actas de las VII Jornadas Nacionales (7º.2022.Bilbao)},

keywords = {Actas, Ciberseguridad, Classification, Darknet Content, VII Jornadas Nacionales},

pubstate = {published},

tppubtype = {article}

}

@article{diaz_optimal_2022,

title = {Optimal botnet detection on network data},

author = {Daniel Díaz and Javier Velasco-Mata and Aitor Del Río Ferreras and Eduardo Fidalgo},

url = {https://dialnet.unirioja.es/servlet/articulo?codigo=9206648},

year  = {2022},

date = {2022-01-01},

journal = {VII Jornadas Nacionales de Investigación en Ciberseguridad 2022},

abstract = {Investigación en Ciberseguridad. Actas de las VII Jornadas Nacionales (JNIC 2022)},

keywords = {JNIC, network data, optimal botnet detection},

pubstate = {published},

tppubtype = {article}

}

@article{diaz_classifying_2022,

title = {Classifying suspicious Pastebin content using Machine Learning},

author = {Daniel Díaz and Francisco Jañez-Martino and Javier Velasco-Mata and Eduardo Fidalgo and Óscar García-Olalla Olivera and Enrique Alegre},

url = {https://dialnet.unirioja.es/servlet/articulo?codigo=9206635},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {VII Jornadas Nacionales de Investigación en Ciberseguridad 2022},

abstract = {Investigación en Ciberseguridad Actas de las VII Jornadas Nacionales (7º.2022.Bilbao) Se enfoca en la construcción de un conjunto de variables para clasificar la maliciosidad de direcciones IP en un entorno de ciberseguridad. Se extrajeron 23 variables, 18 de series temporales y listas de reputación, y 5 relacionadas con la geolocalización. Se realizó un análisis estadístico para evaluar y optimizar las características, considerando la variabilidad temporal de la geolocalización y los hiperparámetros de las series temporales.},

keywords = {Aprendizaje Automático, Ciberseguridad, Direcciones IP, Jornadas Nacionales, Selección de Variables, Severidad},

pubstate = {published},

tppubtype = {article}

}

@article{delgado-sotes_dataset_2022,

title = {Dataset creation and feature extraction for thedetection of fraudulent websites},

author = {Juan José Delgado and Manuel Sánchez-Paniagua and Javier Velasco-Mata and Eduardo Fidalgo and Juan Prieto-Carballal and George Azzopardi},

url = {https://dialnet.unirioja.es/servlet/articulo?codigo=9206641},

year  = {2022},

date = {2022-01-01},

urldate = {2022-01-01},

journal = {Investigación en Ciberseguridad Actas de las VII Jornadas Nacionales (7º. 2022. Bilbao)},

pages = {267–268},

abstract = {Este trabajo se centra en la creación de un conjunto de variables (features) para clasificar la maliciosidad de una dirección IP en un contexto multi-clase. Se han utilizado 23 variables, de las cuales 18 provienen de series temporales y listas de reputación, y 5 están relacionadas con la geolocalización de la IP. Además, se realizó un análisis estadístico para optimizar y estudiar la adecuación de estas variables, teniendo en cuenta los cambios posibles en la geolocalización y los hiperparámetros de las series temporales.},

note = {Publisher: Fundación Tecnalia Research and Innovation},

keywords = {Aprendizaje Automático, Direcciones IP, Selección de Variables, Severidad},

pubstate = {published},

tppubtype = {article}

}

@article{sanchez-paniagua_phishing_2022,

title = {Phishing websites detection using a novel multipurpose dataset and web technologies features},

author = {Manuel Sánchez-Paniagua and Eduardo Fidalgo and Enrique Alegre and Rocío Alaiz-Rodríguez},

url = {https://www.sciencedirect.com/science/article/pii/S0957417422012301},

year  = {2022},

date = {2022-01-01},

journal = {Expert Systems with Applications},

volume = {207},

pages = {118010},

abstract = {Phishing attacks are a major challenge in cybersecurity, often involving the hijacking of sensitive data through fraudulent login forms. This paper proposes a new methodology for detecting phishing websites in real-world scenarios using URL, HTML, and web technology features. The authors introduce the Phishing Index Login Websites Dataset (PILWD), an offline dataset containing 134,000 verified samples, which enables researchers to test and compare detection approaches. Using the dataset, a LightGBM classifier with 54 features achieves a 97.95% accuracy in detecting phishing websites. This methodology is independent of third-party services and utilizes new features for improved detection.},

note = {Publisher: Pergamon},

keywords = {Dataset Creation, LightGBM Classifier, phishing detection, Web Technology Features},

pubstate = {published},

tppubtype = {article}

}

@article{joshi_ranksumunsupervised_2022,

title = {RankSum—An unsupervised extractive text summarization based on rank fusion},

author = {Akanksha Joshi and Eduardo Fidalgo and Enrique Alegre and Rocío Alaiz-Rodríguez},

url = {https://www.sciencedirect.com/science/article/pii/S0957417422002998},

year  = {2022},

date = {2022-01-01},

journal = {Expert Systems with Applications},

volume = {200},

pages = {116846},

abstract = {This paper introduces Ranksum, an approach for extractive text summarization that combines four multi-dimensional sentence features: topic information, semantic content, significant keywords, and position. It ranks sentences based on these features using an unsupervised method, followed by a weighted fusion to determine sentence significance. The method utilizes probabilistic topic models for topic ranking, sentence embeddings for semantic information, and graph-based strategies for identifying keywords. The approach also employs a novelty measure to avoid redundancy. Experimental results on datasets like CNN/DailyMail and DUC 2002 show that Ranksum outperforms existing summarization methods.},

note = {Publisher: Pergamon},

keywords = {Extractive Summarization, Sentence Ranking, Text summarization, Unsupervised Learning},

pubstate = {published},

tppubtype = {article}

}

@article{sanchez-paniagua_phishing_2022-1,

title = {Phishing URL detection: A real-case scenario through login URLs},

author = {Manuel Sánchez-Paniagua and Eduardo Fidalgo and Enrique Alegre and Wesam Al-Nabki and Víctor González-Castro},

url = {https://ieeexplore.ieee.org/abstract/document/9759382},

year  = {2022},

date = {2022-01-01},

journal = {IEEE Access},

volume = {10},

pages = {42949–42960},

abstract = {This paper compares machine learning and deep learning techniques to detect phishing websites through URL analysis. Unlike most current methods, which use only homepages, this study includes URLs from login pages for both legitimate and phishing websites, providing a more realistic scenario. It also demonstrates that existing techniques have high false-positive rates when tested on URLs from legitimate login pages. The authors create a new dataset, Phishing Index Login URL (PILU-90K), and show how older models decrease in accuracy over time. A Logistic Regression model with TF-IDF feature extraction achieves 96.50% accuracy on the login URL dataset.},

note = {Publisher: IEEE},

keywords = {Dataset Creation, machine learning, phishing detection, URL analysis},

pubstate = {published},

tppubtype = {article}

}

@article{bennabhaktula_source_2022,

title = {Source camera device identification from videos},

author = {Guru Swaroop Bennabhaktula and Derrick Timmerman and Enrique Alegre and George Azzopardi},

url = {https://link.springer.com/article/10.1007/s42979-022-01202-0},

year  = {2022},

date = {2022-01-01},

journal = {SN Computer Science},

volume = {3},

number = {4},

pages = {316},

abstract = {This paper addresses the problem of source camera identification in digital videos using deep learning. The authors evaluate different models for camera identification, showing that traditional scene-suppression techniques don't improve performance. They achieved state-of-the-art accuracy on the VISION and QUFVD datasets, outperforming previous methods. The proposed approach does not require flat frames, unlike traditional PRNU-based methods, and is more efficient, making it suitable for use by Law Enforcement Agencies (LEAs).},

note = {Publisher: Springer Nature Singapore Singapore},

keywords = {Camera Recognition, deep learning, Digital Image Forensics, Manchine Learning, Source Camera Identification},

pubstate = {published},

tppubtype = {article}

}

@article{bennabhaktula_camera_2022,

title = {Camera model identification based on forensic traces extracted from homogeneous patches},

author = {Guru Swaroop Bennabhaktula and Enrique Alegre and Dimka Karastoyanova and George Azzopardi},

url = {https://www.sciencedirect.com/science/article/pii/S0957417422010430},

year  = {2022},

date = {2022-01-01},

journal = {Expert Systems with Applications},

volume = {206},

pages = {117769},

abstract = {This work addresses the challenge of identifying the source camera model in digital image forensics, crucial for investigations by Law Enforcement Agencies. The proposed solution extracts small homogeneous regions from the integral image and uses a hierarchical classification approach with convolutional neural networks. This method outperforms traditional classifiers and achieves a 99.01% accuracy on the Dresden dataset’s ‘natural’ subset, marking the best result reported to date.},

note = {Publisher: Pergamon},

keywords = {Camera Model Identification, Digital Image Forensics, Sensor Pattern Noise},

pubstate = {published},

tppubtype = {article}

}

@article{blanco-medina_survey_2022,

title = {A survey on methods, datasets and implementations for scene text spotting},

author = {Pablo Blanco-Medina and Eduardo Fidalgo and Enrique Alegre and Víctor González-Castro},

url = {https://ietresearch.onlinelibrary.wiley.com/doi/full/10.1049/ipr2.12574},

year  = {2022},

date = {2022-01-01},

journal = {IET Image Processing},

volume = {16},

number = {13},

pages = {3426–3445},

abstract = {ext Spotting combines the tasks of detecting and transcribing text present in images, addressing challenges like orientation, aspect ratio, vertical text, and multiple languages in a single image. This paper analyzes and compares the most recent methods and publications in the field, extending beyond traditional comparisons of architectures and performance. It also discusses aspects often overlooked, such as hardware, software, backbone architectures, main challenges, and programming languages used in algorithms. The review covers research from 2016 to 2022, highlighting current problems, future trends, and providing a baseline for the development and comparison of future Text Spotting methods.},

keywords = {Computer vision, image text detection, OCR, text spotting},

pubstate = {published},

tppubtype = {article}

}

@article{jeuland_assessment_2022,

title = {Assessment of age estimation methods for forensic applications using non-occluded and synthetic occluded facial images},

author = {Elouan Derenee Jeuland and Aitor Del Río Ferreras and Deisy Chaves and Eduardo Fidalgo and Víctor González-Castro and Enrique Alegre},

url = {https://ruc.udc.es/dspace/handle/2183/31412},

year  = {2022},

date = {2022-01-01},

journal = {XLIII Jornadas de Automática},

pages = {972–979},

abstract = {This paper evaluates the performance of six deep-learning-based age estimators for forensic applications, particularly in identifying minors and offenders in Child Sexual Exploitation Materials (CSEM). While deep learning is the state-of-the-art for age estimation, it struggles with minors and older adults due to dataset imbalances. Additionally, offenders often use facial occlusion to obscure identities, further impacting estimator accuracy. The study assesses models on non-occluded and synthetically occluded datasets, revealing that eye occlusion has a greater effect than mouth occlusion. Minors and elderly individuals are the most affected by occlusion, making this research a valuable benchmark for forensic victim profiling.},

note = {Publisher: Universidade da Coruña. Servizo de Publicacións},

keywords = {Age Estimation, CSEM, deep learning, facial occlusion},

pubstate = {published},

tppubtype = {article}

}

@article{riego_del_castillo_estimation_2022,

title = {Estimation of lamb weight using transfer learning and regression},

author = {Virginia Riego del Castillo and Lidia Sánchez-González and Laura Fernández-Robles and Manuel Castejón-Limas and Rubén Rebollar},

url = {https://link.springer.com/chapter/10.1007/978-3-031-18050-7_3},

year  = {2022},

date = {2022-01-01},

journal = {International Workshop on Soft Computing Models in Industrial and Environmental Applications},

pages = {23–30},

abstract = {This paper proposes an automatic, non-invasive, and cost-effective method to estimate the weight of live lambs using a camera, such as those found in mobile phones. The approach employs a pre-trained Convolutional Neural Network (Xception) with transfer learning to estimate lamb weight based on an image, the lamb's sex, and the height from which the image is taken. The method achieved a mean absolute error (MAE) of 0.58 kg and an R² value of 0.96, offering improved accuracy over traditional methods, and providing a practical solution to estimate lamb weight with minimal input and error.},

note = {Publisher: Springer Nature Switzerland Cham},

keywords = {convolutional neural network, Lamb Weight Estimation, Non-invasive Measurement, Transfer Learning},

pubstate = {published},

tppubtype = {article}

}

@article{fernandez-robles_analytical_2022,

title = {Analytical Framework to Investigate Ethics, Social Responsibility and Sustainability in Engineering Project Management},

author = {Laura Fernández-Robles and Manuel Castejón-Limas and Alexis Gutiérrez-Fernández and Francisco J Rodríguez-Lera and Camino Fernández-Llamas},

url = {https://scholar.google.com/scholar?cluster=3250378527849384264&hl=en&inst=7489717285800386075&inst=6102321801771322873&oi=scholarr},

year  = {2022},

date = {2022-01-01},

journal = {INTERNATIONAL JOURNAL OF ENGINEERING EDUCATION},

volume = {38},

number = {3},

pages = {673–683},

abstract = {Project management is a key part of engineering curricula worldwide, often focused on balancing quality, cost, and schedule. However, ethics, social responsibility, and sustainability are also crucial for project success, as they ensure compliance with laws, regulations, and local values. This paper analyzes how ethics, social responsibility, and sustainability are addressed in two widely-used project management standards: the Project Management Book of Knowledge (PMBoK) and the Individual Competence Baseline for Project, Program & Portfolio Management (ICB). An analytical framework is designed for a desk research of these standards.},

note = {Publisher: TEMPUS PUBLICATIONS IJEE, ROSSMORE,, DURRUS, BANTRY, COUNTY CORK 00000, IRELAND},

keywords = {ethics, PMBoK, project management, social responsibility, sustainability},

pubstate = {published},

tppubtype = {article}

}

@article{castejon-limas_framework_2022,

title = {A framework for the optimization of complex cyber-physical systems via directed acyclic graph},

author = {Manuel Castejón-Limas and Laura Fernández-Robles and Héctor Alaiz-Moretón and Jaime Cifuentes-Rodriguez and Camino Fernández-Llamas},

url = {https://www.mdpi.com/1424-8220/22/4/1490},

year  = {2022},

date = {2022-01-01},

journal = {Sensors},

volume = {22},

number = {4},

pages = {1490},

abstract = {Mathematical modeling and data-driven methodologies are frequently required to optimize industrial processes in the context of Cyber-Physical Systems (CPS). This paper introduces the PipeGraph software library, an open-source python toolbox for easing the creation of machine learning models by using Directed Acyclic Graph (DAG)-like implementations that can be used for CPS. scikit-learn’s Pipeline is a very useful tool to bind a sequence of transformers and a final estimator in a single unit capable of working itself as an estimator. It sequentially assembles several steps that can be cross-validated together while setting different parameters. Steps encapsulation secures the experiment from data leakage during the training phase. The scientific goal of PipeGraph is to extend the concept of Pipeline by using a graph structure that can handle scikit-learn’s objects in DAG layouts. It allows performing diverse operations, instead of only transformations, following the topological ordering of the steps in the graph; it provides access to all the data generated along the intermediate steps; and it is compatible with GridSearchCV function to tune the hyperparameters of the steps. It is also not limited to (𝑋,𝑦) entries. Moreover, it has been proposed as part of the scikit-learn-contrib supported project, and is fully compatible with scikit-learn. Documentation and unitary tests are publicly available together with the source code. Two case studies are analyzed in which PipeGraph proves to be essential in improving CPS modeling and optimization: the first is about the optimization of a heat exchange management system, and the second deals with the detection of anomalies in manufacturing processes.},

note = {Publisher: MDPI},

keywords = {Cyber-Physical Systems, Directed Acyclic Graphs, Lean Manufacturing, machine learning models, pipegraph, scikit-learn},

pubstate = {published},

tppubtype = {article}

}

@article{castano_state_2021,

title = {State of the art: content-based and hybrid phishing detection},

author = {Felipe Castano and Eduardo Fidalgo and Enrique Alegre and Deisy Chaves and Manuel Sánchez-Paniagua},

url = {https://arxiv.org/abs/2101.12723},

year  = {2021},

date = {2021-01-01},

journal = {arXiv preprint arXiv:2101.12723},

abstract = {Phishing attacks have evolved and increased over time and, for this reason, the task of distinguishing between a legitimate site and a phishing site is more and more difficult, fooling even the most expert users. The main proposals focused on addressing this problem can be divided into four approaches: List-based, URL based, content-based, and hybrid. In this state of the art, the most recent techniques using web content-based and hybrid approaches for Phishing Detection are reviewed and compared.},

keywords = {Cybersecurity, Hybrid Phishing, phishing detection},

pubstate = {published},

tppubtype = {article}

}

@article{saikia_colour_2021,

title = {Colour Neural Descriptors for Instance Retrieval Using CNN Features and Colour Models},

author = {Surajit Saikia and Laura Fernández-Robles and Eduardo Fidalgo and Enrique Alegre},

url = {https://ieeexplore.ieee.org/abstract/document/9344701},

year  = {2021},

date = {2021-01-01},

journal = {IEEE Access},

volume = {9},

pages = {23218–23234},

abstract = {This paper presents color neural descriptors for image retrieval, using CNN features from different color spaces without fine-tuning. An object detector enhances feature extraction, and a stride-based query expansion improves multi-view retrieval. The method achieves state-of-the-art results on multiple datasets.},

note = {Publisher: IEEE},

keywords = {CNN Features, Color Descriptors, deep learning, Image Retrieval, Object Detection},

pubstate = {published},

tppubtype = {article}

}

@article{gangwar_attm-cnn_2021,

title = {AttM-CNN: Attention and metric learning based CNN for pornography, age and Child Sexual Abuse (CSA) Detection in images},

author = {Abhishek Gangwar and Víctor González-Castro and Enrique Alegre and Eduardo Fidalgo},

url = {https://www.sciencedirect.com/science/article/pii/S092523122100312X},

year  = {2021},

date = {2021-01-01},

journal = {Neurocomputing},

volume = {445},

pages = {81–104},

abstract = {This paper proposes AttM-CNN, a deep learning model for detecting Child Sexual Abuse (CSA) material by combining pornographic content detection and age-group classification. Two new datasets, Pornographic-2M and Juvenile-80k, are introduced for training. The model outperforms state-of-the-art methods and improves CSA detection accuracy over forensic tools, aiding law enforcement.},

note = {Publisher: Elsevier},

keywords = {age-group detection, CNN, convolutional neural network, CSA detection, metric learning, pornography detection, visual attention},

pubstate = {published},

tppubtype = {article}

}

@article{janez-martino_trustworthiness_2021,

title = {Trustworthiness of spam email addresses using machine learning},

author = {Francisco Jáñez-Martino and Rocío Alaiz-Rodríguez and Víctor González-Castro and Eduardo Fidalgo},

url = {https://dl.acm.org/doi/abs/10.1145/3469096.3475060},

year  = {2021},

date = {2021-01-01},

journal = {Proceedings of the 21st ACM Symposium on Document Engineering},

pages = {1–4},

abstract = {This paper addresses the growing issue of spam emails used by cybercriminals for scams, phishing, and malware attacks. It presents a proof-of-concept methodology to help users assess the trustworthiness of email addresses. The authors introduce a manually labeled dataset of email addresses, categorized as low and high quality, and extract 18 handcrafted features based on social engineering techniques and natural language properties. Four machine learning classifiers are tested, with Naive Bayes yielding the best performance (88.17% accuracy and 0.808 F1-Score). The study also utilizes the InterpretML framework to identify the most relevant features for building an automatic system to assess email address trustworthiness.},

keywords = {Cybersecurity, machine learning, Phishing, Spam Email Detection, Trustworthiness Analysis},

pubstate = {published},

tppubtype = {article}

}

@article{cueto-lopez_evaluation_2021,

title = {Evaluation of feature selection techniques for breast cancer risk prediction},

author = {Nahúm Cueto-López and María Teresa García-Ordás and Facundo Vitelli-Storelli and Pablo Fernández-Navarro and Camilo Palazuelos and Rocío Alaiz-Rodríguez},

url = {https://www.mdpi.com/1660-4601/18/20/10670},

year  = {2021},

date = {2021-01-01},

journal = {International Journal of Environmental Research and Public Health},

volume = {18},

number = {20},

pages = {10670},

abstract = {This study evaluates feature selection techniques combined with machine learning classifiers to improve breast cancer risk prediction. Using data from the MCC-Spain study (919 cases, 946 controls) with environmental and genetic features, the goal is to identify key risk factors and assess the stability of feature selection methods. SVM-RFE achieved the best performance, with a Logistic Regression model using its top-47 ranked features obtaining an AUC of 0.616 (5.8% improvement). SVM-RFE and Random Forest were the most stable selection methods, but SVM-RFE outperformed Random Forest in predictive accuracy.},

note = {Publisher: MDPI},

keywords = {Breast Cancer, feature selection, Risk Prediction Model, stability},

pubstate = {published},

tppubtype = {article}

}

@article{biswas_new_2021,

title = {A new perceptual hashing method for verification and identity classification of occluded faces},

author = {Rubel Biswas and Víctor González-Castro and Eduardo Fidalgo and Enrique Alegre},

url = {https://www.sciencedirect.com/science/article/pii/S0262885621001505},

year  = {2021},

date = {2021-01-01},

journal = {Image and Vision Computing},

volume = {113},

pages = {104245},

abstract = {Recently, research communities on Computer Vision and biometrics have shown a lot of interest in face verification and classification methods. Fighting against Child Sexual Exploitation Material (CSEM) is one of the applications that might benefit most from these advances. In CSEM, discriminative parts of the face, i.e. mostly the eyes, are often occluded to make the victim identification more difficult. Most of the current face recognition methods are not able to handle such kind of occlusions. To overcome this problem, we present One-Shot Frequency Dominant Neighborhood Structure (OSF-DNS), a new perceptual hashing method that shows advantages on two scenarios: (a) occluded face verification, i.e., matching occluded faces with their non-occluded versions, and (b) face classification, i.e., getting the identity of an occluded face by means of a classifier trained with the non-occluded faces using the perceptual hash codes as feature vectors. We have compared the face verification performance of OSF-DNS with three perceptual hashing methods and with the features obtained from five deep learning techniques, using the occluded versions of six different facial datasets. The proposed method achieves accuracies between 69.24% and 99.46% depending on the dataset, and always higher than the compared methods. For the face classification task, we compared the performance of OSF-DNS with the features obtained by four deep learning techniques. Experimental results on LFW and CFPW datasets showed that the proposed hashing method outperformed the results obtained with deep features with an accuracy up to 89.53%.},

note = {Publisher: Elsevier},

keywords = {biometrics, face verification, oclussion handling, perceptual hashing},

pubstate = {published},

tppubtype = {article}

}

@article{biswas_video_2021,

title = {A video summarization approach to speed-up the analysis of child sexual exploitation material},

author = {Rubel Biswas and Deisy Chaves and Laura Fernández-Robles and Eduardo Fidalgo and Enrique Alegre},

url = {https://ruc.udc.es/dspace/handle/2183/28353},

year  = {2021},

date = {2021-01-01},

journal = {XLII Jornadas de Automática},

pages = {648–654},

abstract = {This paper presents a video summarization strategy combining perceptual hashing and face detection to identify key frames from videos, specifically targeting content with faces that may relate to victims or offenders. The proposed approach is tested on adult pornography detection using the NDPI-800 dataset, achieving 84.15% accuracy and a speed of 8.05 ms/frame, making it suitable for real-time applications. This method can also create video summaries while preserving distinctive faces from the original footage.},

note = {Publisher: Universidade da Coruña, Servizo de Publicacións},

keywords = {content detection, face detection, perceptual hashing, real time applications, video summarization},

pubstate = {published},

tppubtype = {article}

}

@article{sanchez-paniagua_fraudulent_2021,

title = {Fraudulent e-commerce websites detection through machine learning},

author = {Manuel Sánchez-Paniagua and Eduardo Fidalgo and Enrique Alegre and Francisco Jáñez-Martino},

url = {https://link.springer.com/chapter/10.1007/978-3-030-86271-8_23},

year  = {2021},

date = {2021-01-01},

journal = {Hybrid Artificial Intelligent Systems: 16th International Conference, HAIS 2021, Bilbao, Spain, September 22–24, 2021, Proceedings 16},

pages = {267–279},

abstract = {With the rise of e-commerce, users are increasingly vulnerable to fraudulent websites that sell counterfeit products or steal personal information. Existing protection methods, such as blacklists and rules, are prone to high false-positive rates and require constant updating. This paper presents a publicly available dataset of potentially fraudulent websites, incorporating seven new features for better detection. The model, using Random Forest and 11 handcrafted features, achieved an F1-Score of X on a dataset of 282 samples.},

note = {Publisher: Springer International Publishing},

keywords = {Cybersecurity, E-commerce, Fraud Detection, machine learning},

pubstate = {published},

tppubtype = {article}

}

@article{riego_strong_2021,

title = {Strong classification system for wear identification on milling processes using computer vision and ensemble learning},

author = {Virginia Riego and Manuel Castejón-Limas and Lidia Sánchez-González and Laura Fernández-Robles and Hilde Perez and Javier Díez-González and Ángel-Manuel Guerrero-Higueras},

url = {https://www.sciencedirect.com/science/article/pii/S0925231220316155},

year  = {2021},

date = {2021-01-01},

journal = {Neurocomputing},

volume = {456},

pages = {678–684},

abstract = {This paper proposes a machine-vision-based method for evaluating the texture of the inner and outer surfaces of machined metallic pieces with cylindrical holes. Using a microscope camera connected to a rigid industrial boroscope, images of the hole surface are captured. The texture descriptors extracted from these images are analyzed, and a significant correlation is found. Feature reduction is performed, followed by classification using various algorithms with exhaustive grid search and 10-fold cross-validation. The best results are obtained with the Extremely Randomized Trees classifier, achieving a mean test score of 92.98%, surpassing previous research and meeting industry requirements.},

note = {Publisher: Elsevier},

keywords = {Haralick Descriptors, Milling Machined Parts, Quality Estimation, Wear Detection},

pubstate = {published},

tppubtype = {article}

}

@article{riego_burr_2021,

title = {Burr detection and classification using rustico and image processing},

author = {Virginia Riego and Lidia Sánchez-González and Laura Fernández-Robles and Alexis Gutiérrez-Fernández and Nicola Strisciuglio},

url = {https://www.sciencedirect.com/science/article/pii/S1877750321001538},

year  = {2021},

date = {2021-01-01},

journal = {Journal of computational science},

volume = {56},

pages = {101485},

abstract = {This study focuses on classifying burrs in edge finishing of machined workpieces to reduce production costs and time. It identifies three types of burrs: knife-type (no imperfections), saw-type (small splinters), and burr-breakage (substantial deformation). The proposed method, RUSTICO, automatically classifies the edge of each piece, achieving a 91.2% F1-Score and successfully identifying the burr-breakage type.},

note = {Publisher: Elsevier},

keywords = {Burr Classification, Burrs in Workpiece, Milling Machined Parts, RUSTICO},

pubstate = {published},

tppubtype = {article}

}

@article{fernandez-robles_use_2021,

title = {Use of image processing to monitor tool wear in micro milling},

author = {Laura Fernández-Robles and Lidia Sánchez-González and Javier Díez-González and Manuel Castejón-Limas and Hilde Pérez},

url = {https://www.sciencedirect.com/science/article/pii/S0925231220317501},

year  = {2021},

date = {2021-01-01},

journal = {Neurocomputing},

volume = {452},

pages = {333–340},

abstract = {This paper presents a digital image processing method for monitoring tool wear in micro milling, a process where tools wear quickly due to the small size and complex geometries of the machined components. Direct measurement of tool wear is not feasible due to the size of the tools, so this method uses captured images of the tool to analyze wear progression. The wear is measured in terms of flank wear, crater wear, and tool radius reduction. Several approaches, including morphological operations, k-means clustering, and the Otsu Multilevel algorithm, were compared to determine the best method for analyzing images. The results show a 5% difference between predicted and actual worn areas, meeting industrial standards. This method can be applied in industrial environments and used in collaborative robots to enhance automation and decision-making processes.},

note = {Publisher: Elsevier},

keywords = {digital image processing, Micro Milling, Tool Breakage, Tool wear},

pubstate = {published},

tppubtype = {article}

}

@article{fernandez-raga_use_2021,

title = {The use of high-speed cameras as a tool for the characterization of raindrops in splash laboratory studies},

author = {María Fernández-Raga and Marco Cabeza-Ortega and Víctor González-Castro and Piet Peters and Meindert Commelin and Julián Campo},

url = {https://www.mdpi.com/2073-4441/13/20/2851},

year  = {2021},

date = {2021-01-01},

journal = {Water},

volume = {13},

number = {20},

pages = {2851},

abstract = {This study explores raindrop characterization using high-speed cameras and image processing for rainfall simulation calibration. Two phases were conducted: first, measuring and analyzing individual drops’ speed and shape; second, designing a calibration procedure using multidrop images to assess rain simulators. Results showed that drop shape varies with size, from round to oval, and terminal velocity was accurately measured. It was observed that higher simulated rain intensity produced smaller drops, contrasting with natural rainfall. This calibration helps evaluate rain simulators' realism, determining their kinetic energy and applicability for environmental modeling.},

note = {Publisher: MDPI},

keywords = {High-Speed Camera, Raindrop, Rainfall Simulator, Splash Erosion, Terminal Velocity},

pubstate = {published},

tppubtype = {article}

}

@article{saikia_image_2021,

title = {Image retrieval based on texture using latent space representation of discrete Fourier transformed maps},

author = {Surajit Saikia and Laura Fernández-Robles and Enrique Alegre and Eduardo Fidalgo},

url = {https://link.springer.com/article/10.1007/s00521-021-05955-2},

year  = {2021},

date = {2021-01-01},

journal = {Neural Computing and Applications},

volume = {33},

number = {20},

pages = {13301–13316},

abstract = {This paper presents a Fourier-based approach to texture-based image retrieval, applied to indoor scene images that typically contain multiple texture patterns, making it more challenging compared to traditional fabric or textile retrieval tasks. The proposed method, useful in crime scene investigations for evidence matching, combines spatial images with their discrete Fourier transform maps to create a new texture representation. The framework integrates region proposal networks, convolutional autoencoders, and transfer learning to extract texture descriptors from the encoder's latent space layer. The experimental results on four datasets (TextileTube, Outex, USPtex, and Stex) show that the proposed method outperforms existing techniques.},

note = {Publisher: Springer London},

keywords = {Convolutional Autoencoders, Fourier Transform, Region Proposal Networks, Texture-based Image Retrieval},

pubstate = {published},

tppubtype = {article}

}

@article{castano_evaluation_2021,

title = {Evaluation of state-of-art phishing detection strategies based on machine learning},

author = {Felipe Castaño and Manuel Sánchez-Paniagua and J Delgado and Javier Velasco-Mata and A Sepúlveda and Eduardo Fidalgo and Enrique Alegre},

url = {https://scholar.google.es/citations?view_op=view_citation&hl=es&user=yATJZvcAAAAJ&cstart=20&pagesize=80&sortby=title&citation_for_view=yATJZvcAAAAJ:Tiz5es2fbqcC},

year  = {2021},

date = {2021-01-01},

urldate = {2021-01-01},

journal = {Investigación en Ciberseguridad (Castilla-La Mancha). Ediciones de la Universidad De Castilla-La Mancha},

abstract = {This paper reviews and evaluates current state-of-the-art phishing detection strategies that use machine learning.},

keywords = {Cybersecurity, machine learning, phishing detection},

pubstate = {published},

tppubtype = {article}

}

@article{velasco-mata_efficient_2021,

title = {Efficient detection of botnet traffic by features selection and decision trees},

author = {Javier Velasco-Mata and Víctor González-Castro and Eduardo Fidalgo and Enrique Alegre},

url = {https://ieeexplore.ieee.org/abstract/document/9523853},

year  = {2021},

date = {2021-01-01},

journal = {IEEE Access},

volume = {9},

pages = {120567–120579},

abstract = {Botnets pose a major online threat, causing significant economic losses. With the rise of connected devices, analyzing large network traffic data is crucial. This study enhances botnet traffic classification by selecting the most relevant features using Information Gain and Gini Importance. Three feature subsets (5, 6, and 7 features) were tested with Decision Tree, Random Forest, and k-Nearest Neighbors on two datasets derived from CTU-13 (QB-CTU13 and EQB-CTU13). Results show that Decision Trees with a five-feature set achieved the best performance, with an 85% F1 score and an average classification time of 0.78 microseconds per sample.},

note = {Publisher: IEEE},

keywords = {Botnet Detection, Cybersecurity, feature selection, machine learning, Network Traffic Analysis},

pubstate = {published},

tppubtype = {article}

}

@article{fidalgo_state_2021,

title = {State of the Art: Content-based and Hybrid Phishing},

author = {Felipe Castaño and Eduardo Fidalgo and Enrique Alegre and Deisy Chaves and Manuel Sánchez-Paniagua},

url = {https://arxiv.org/abs/2101.12723},

year  = {2021},

date = {2021-01-01},

urldate = {2021-01-01},

abstract = {Phishing attacks have evolved and increased over time and, for this reason, the task of distinguishing between a legitimate site and a phishing site is more and more difficult, fooling even the most expert users. The main proposals focused on addressing this problem can be divided into four approaches: List-based, URL based, content-based, and hybrid. In this state of the art, the most recent techniques using web content-based and hybrid approaches for Phishing Detection are reviewed and compared.},

keywords = {Content-based Features, Cybersecurity, deep learning, Hybrid Features, Hybrid Phishing, machine learning, phishing detection},

pubstate = {published},

tppubtype = {article}

}

@article{blanco-medina_detecting_2021,

title = {Detecting vulnerabilities in critical infrastructures by classifying exposed industrial control systems using deep learning},

author = {Pablo Blanco-Medina and Eduardo Fidalgo and Enrique Alegre and Andrés Carofilis-Vasco and Francisco Jáñez-Martino and Víctor Fidalgo-Villar},

url = {https://www.mdpi.com/2076-3417/11/1/367},

year  = {2021},

date = {2021-01-01},

urldate = {2021-01-01},

journal = {Applied Sciences},

volume = {11},

number = {1},

pages = {367},

abstract = {This paper presents a deep learning pipeline to classify industrial control panel screenshots into three categories: internet technologies, operation technologies, and others. Using the CRINF-300 dataset, the authors compared CNN architectures and found that Inception-ResNet-V2 and VGG16 performed best, while MobileNet-V1 was recommended for time-sensitive systems with GPU availability.},

note = {Publisher: MDPI},

keywords = {deep learning, Fine-tuning, Image classification, Industrial Control System, Transfer Learning},

pubstate = {published},

tppubtype = {article}

}

@article{molpeceres-barrientos_machine_2020,

title = {Machine learning techniques for the detection of inappropriate erotic content in text},

author = {Gonzalo Molpeceres-Barrientos and Rocío Alaiz-Rodríguez and Víctor González-Castro and Andrew Parnell},

url = {https://link.springer.com/article/10.2991/ijcis.d.200519.003},

year  = {2020},

date = {2020-01-01},

urldate = {2020-01-01},

journal = {International Journal of Computational Intelligence Systems},

volume = {13},

number = {1},

pages = {591–603},

abstract = {This study addresses the problem of detecting erotic or sexual content in text documents, specifically for protecting children online. Using Natural Language Processing (NLP) techniques, the authors evaluated twelve models combining different text encoders (Bag of Words, TF-IDF, and Word2vec) with various classifiers (SVM, Logistic Regression, k-NN, and Random Forest). The evaluation was conducted on a dataset created from Reddit. The best result was achieved using TF-IDF with an SVM classifier, which achieved an accuracy of 0.97 and an F-score of 0.96 (precision 0.96/recall 0.95). This demonstrates the feasibility of detecting erotic content and creating filters for minors or user preferences.},

note = {Publisher: Springer Netherlands Dordrecht},

keywords = {machine learning, Natural Language Processing, NLP, Text classification},

pubstate = {published},

tppubtype = {article}

}

@article{gonzalez-castro_ct_2020,

title = {CT radiomics in colorectal cancer: Detection of KRAS mutation using texture analysis and machine learning},

author = {Víctor González-Castro and Eva Cernadas and Emilio Huelga and Manuel Fernández-Delgado and Jacobo Porto-Álvarez and José Ramón Antúnez-López and Miguel Souto-Bayarri},

url = {https://www.mdpi.com/2076-3417/10/18/6214},

year  = {2020},

date = {2020-01-01},

urldate = {2020-01-01},

journal = {Applied Sciences},

volume = {10},

number = {18},

pages = {6214},

abstract = {This study analyzes CT image textures to classify colorectal cancer patients as KRAS+ or KRAS- using machine learning classifiers (SVM, GBM, NNET, RF). Texture analysis quantifies tumor heterogeneity, supporting the use of radiomics to predict KRAS mutations. A retrospective study with 47 patients achieved the highest accuracy (83%) and kappa (64.7%) using NNET with wavelet transform and Haralick features. This non-invasive approach could eliminate the need for biopsies, reducing risks and enabling more personalized treatment.},

note = {Publisher: MDPI},

keywords = {colorectal cancer, KRAS Mutation, machine learning, Non-invasive Diagnosis, Radiomics},

pubstate = {published},

tppubtype = {article}

}

@article{al-nabki_file_2020,

title = {File Name Classification Approach to Identify Child Sexual Abuse},

author = {Wesam Al-Nabki and Eduardo Fidalgo and Enrique Alegre and Rocío Alaiz-Rodríguez},

url = {https://www.scitepress.org/PublishedPapers/2020/91548/91548.pdf},

year  = {2020},

date = {2020-01-01},

journal = {Proceedings of the 9th International Conference on Pattern Recognition Applications and Methods},

volume = {1},

number = {978-989-758-397-1},

abstract = {This study explores automatic methods to help law enforcement detect Child Sexual Exploitation Material (CSEM) by analyzing file names instead of inspecting visual content. It compares traditional machine learning models (Logistic Regression, SVM) with deep learning approaches using character-level CNNs. The CNN model showed promising results, achieving a 0.86 average recall and 0.78 for the CSEM class, making it a useful tool for time-constrained forensic investigations.},

keywords = {Active Learning, Character-level Convolutional Networks, Child Sexual Abuse, File Name Classification, Short Text Classification},

pubstate = {published},

tppubtype = {article}

}

@article{chaves_improving_2020,

title = {Improving age estimation in minors with occluded faces to fight against child sexual exploitation},

author = {Deisy Chaves and Eduardo Fidalgo and Enrique Alegre and Francisco Jañez-Martino and Rubel Biswas},

url = {https://scholar.google.es/citations?view_op=view_citation&hl=es&user=yATJZvcAAAAJ&cstart=20&pagesize=80&sortby=title&citation_for_view=yATJZvcAAAAJ:35N4QoGY0k4C},

year  = {2020},

date = {2020-01-01},

urldate = {2020-01-01},

journal = {Proceedings of the 15th Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications},

volume = {1},

number = {978-989-758-402-2},

abstract = {Proceedings of the 15th Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications},

keywords = {Age Estimation, facial occlusion},

pubstate = {published},

tppubtype = {article}

}

@article{fidalgo_classifying_2020,

title = {Classifying Suspicious Content in Tor Darknet},

author = {Eduardo Fidalgo and Andrés Carofilis-Vasco and Francisco Jáñez-Martino and Pablo Blanco-Medina},

url = {https://arxiv.org/abs/2005.10086},

year  = {2020},

date = {2020-01-01},

journal = {arXiv e-prints},

pages = {arXiv–2005},

abstract = {This paper proposes Semantic Attention Keypoint Filtering (SAKF) to classify Tor Darknet images by focusing on significant features related to criminal activities. By combining saliency maps with Bag of Visual Words (BoVW), SAKF outperforms CNN approaches (MobileNet v1, ResNet50) and BoVW with dense SIFT descriptors, achieving 87.98% accuracy.},

keywords = {Computer vision, Criminal Activity Detection, Darknet Analysis, Image classification},

pubstate = {published},

tppubtype = {article}

}

@article{sanchez-paniagua_impact_2020,

title = {Impact of current phishing strategies in machine learning models for phishing detection},

author = {Manuel Sánchez-Paniagua and Eduardo Fidalgo and Víctor González-Castro and Enrique Alegre},

url = {https://link.springer.com/chapter/10.1007/978-3-030-57805-3_9},

year  = {2020},

date = {2020-01-01},

journal = {13th International Conference on Computational Intelligence in Security for Information Systems (CISIS)},

pages = {87–96},

abstract = {Phishing is one of the most widespread attacks based on social engineering. The detection of Phishing using Machine Learning approaches is more robust than the blacklist-based ones, which need regular reports and updates. However, the datasets currently used for training the Supervised Learning approaches have some drawbacks. These datasets only have the landing page of legitimate domains and they do not include the login forms from the websites, which is the most common situation in a real case of Phishing. This makes the performance of Machine Learning-based models to drop, especially when they are tested using login pages.},

keywords = {machine learning, NLP, phishing detection, URL},

pubstate = {published},

tppubtype = {article}

}

@article{fidalgo_classifying_2020-1,

title = {Classifying suspicious content in Tor Darknet},

author = {Eduardo Fidalgo and Andrés Carofilis-Vasco and Francisco Jáñez-Martino and Pablo Blanco-Medina},

url = {https://ui.adsabs.harvard.edu/abs/2020arXiv200510086F/abstract},

year  = {2020},

date = {2020-01-01},

journal = {arXiv preprint arXiv:2005.10086},

abstract = {This paper proposes Semantic Attention Keypoint Filtering (SAKF) to classify Tor Darknet images by focusing on significant features related to criminal activities. By combining saliency maps with Bag of Visual Words (BoVW), SAKF outperforms CNN approaches (MobileNet v1, ResNet50) and BoVW with dense SIFT descriptors, achieving 87.98% accuracy.},

keywords = {Computer vision, Criminal Activity Detection, Darknet Analysis, Image classification},

pubstate = {published},

tppubtype = {article}

}

@article{al-nabki_improving_2020,

title = {Improving named entity recognition in tor darknet with local distance neighbor feature},

author = {Wesam Al-Nabki and Francisco Jáñez-Martino and Andrés Carofilis-Vasco and Eduardo Fidalgo and Javier Velasco-Mata},

url = {https://arxiv.org/abs/2005.08746},

year  = {2020},

date = {2020-01-01},

urldate = {2020-01-01},

journal = {arXiv preprint arXiv:2005.08746},

abstract = {This paper introduces a novel feature called Local Distance Neighbor (LDN) for named entity recognition in noisy user-generated texts, replacing the need for task-specific and costly gazetteers. The approach was tested on the W-NUT-2017 dataset, achieving state-of-the-art results for Group, Person, and Product categories. By adding 851 manually labeled samples, the method also demonstrated effectiveness in detecting named entities in the Tor Darknet, with F1 scores of 52.96% and 50.57%, aiding Law Enforcement Agencies in identifying entities related to weapons and drug selling.},

keywords = {Gazetteer, NER, Tor Darknet},

pubstate = {published},

tppubtype = {article}

}

@article{al-nabki_evaluating_2020,

title = {Evaluating performance of an adult pornography classifier for child sexual abuse detection},

author = {Wesam Al-Nabki and Eduardo Fidalgo and Andrés Carofilis-Vasco and Francisco Jáñez-Martino and Javier Velasco-Mata},

url = {https://arxiv.org/abs/2005.08766},

year  = {2020},

date = {2020-01-01},

journal = {arXiv preprint arXiv:2005.08766},

abstract = {The IT revolution has increased access to pornographic content, raising concerns about child abuse detection. This study evaluates the hardware and software factors affecting forensic tools that classify adult content. The Yahoo Deep Learning-based classifier was tested on different OS and hardware setups. Results show that Ubuntu outperforms Windows 10, being 5× faster on CPU and 2× faster on GPU. GPU-based machines significantly outperform CPU-based ones (7–8× faster). Image resizing (interpolation) has no impact on model performance.},

keywords = {Deep Learning Classification, Forensic Tools, Hardware and OS Performance, Pornographic Content Detection},

pubstate = {published},

tppubtype = {article}

}

@article{janez-martino_classification_2020,

title = {Classification of spam emails through hierarchical clustering and supervised learning},

author = {Francisco Jánez-Martino and Eduardo Fidalgo and Santiago González and Javier Velasco-Mata},

url = {https://arxiv.org/abs/2005.08773},

year  = {2020},

date = {2020-01-01},

urldate = {2020-01-01},

journal = {arXiv preprint arXiv:2005.08773},

abstract = {This work introduces SPEMC-11K, the first multi-class spam email dataset, categorizing spam into Health and Technology, Personal Scams, and Sexual Content. Using TF-IDF and BOW with Naïve Bayes, Decision Trees, and SVM, the best accuracy (95.39% F1-score) is achieved with TF-IDF and SVM, while TF-IDF and NB offer the fastest classification (2.13ms per email).},

keywords = {Cybersecurity, machine learning, Spam Classification, Text Processing, TF-IDF & BOW},

pubstate = {published},

tppubtype = {article}

}

@article{biswas_perceptual_2020-1,

title = {Perceptual Hashing applied to Tor domains recognition},

author = {Rubel Biswas and Andrés Carofilis-Vasco and Eduardo Fidalgo and Francisco Jáñez-Martino and Pablo Blanco-Medina},

url = {https://arxiv.org/abs/2005.10090},

year  = {2020},

date = {2020-01-01},

urldate = {2020-01-01},

journal = {arXiv preprint arXiv:2005.10090},

abstract = {This paper introduces Frequency-Dominant Neighborhood Structure (F-DNS), a perceptual hashing method for automatically classifying Tor domains by their screenshots. F-DNS outperforms other methods, achieving better correlation coefficients, especially for rotated images. The method was tested on the Darknet Usage Service Images-2K (DUSI-2K) dataset and achieved an accuracy of 98.75%, surpassing other classification and hashing techniques.},

keywords = {Cybersecurity, DCT, Deep Web, Image classification, TOR},

pubstate = {published},

tppubtype = {article}

}

@article{biswas_perceptual_2020,

title = {Perceptual image hashing based on frequency dominant neighborhood structure applied to Tor domains recognition},

author = {Rubel Biswas and Víctor González-Castro and Eduardo Fidalgo and Enrique Alegre},

url = {https://www.sciencedirect.com/science/article/pii/S0925231219316674},

year  = {2020},

date = {2020-01-01},

journal = {Neurocomputing},

volume = {383},

pages = {24–38},

abstract = {This paper proposes an automatic method to recognize illicit domains on the Tor network using perceptual hashing through domain snapshots. The method introduces DUSI-2K, a dataset of Tor service domain snapshots, and F-DNS, a new hashing technique based on Dominant Neighborhood Structure (DNS) and Global Neighborhood Structure (GNS). F-DNS outperforms other state-of-the-art methods, achieving an accuracy of 98.75% in recognizing Tor domains, significantly surpassing methods like ResNet50 and Inception-ResNet-v2. Fine-tuning these models does not improve results, demonstrating the effectiveness of F-DNS for Tor domain classification.},

note = {Publisher: Elsevier},

keywords = {Cybersecurity, Deep Web, perceptual hashing, TOR},

pubstate = {published},

tppubtype = {article}

}

@article{bennabhaktula_device-based_2020,

title = {Device-based image matching with similarity learning by convolutional neural networks that exploit the underlying camera sensor pattern noise},

author = {Guru Swaroop Bennabhaktula and Enrique Alegre and Dimka Karastoyanova and George Azzopardi},

url = {https://arxiv.org/abs/2004.11443},

year  = {2020},

date = {2020-01-01},

journal = {arXiv preprint arXiv:2004.11443},

abstract = {This paper addresses the challenge of identifying whether two images originate from the same camera, aiding forensic investigations. A two-part network is proposed to quantify the likelihood of a shared source, evaluated on the Dresden dataset (1851 images from 31 cameras). While not yet forensics-ready, the approach achieves 85% accuracy, showing promising results. This research is part of the EU-funded 4NSEEK project focused on combating child sexual abuse.},

keywords = {4NSEEK project, Camera Identification, deep learning, DigitalImage Forensics},

pubstate = {published},

tppubtype = {article}

}

@article{blanco-medina_classification_2020,

title = {Classification of Industrial Control Systems screenshots using Transfer Learning},

author = {Pablo Blanco-Medina and Eduardo Fidalgo and Enrique Alegre and Francisco Jáñez-Martino and Andrés Carofilis-Vasco and Víctor Fidalgo-Villar},

year  = {2020},

date = {2020-01-01},

journal = {arXiv e-prints},

pages = {arXiv–2005},

abstract = {This study evaluates CNN-based transfer learning for classifying Industrial Control System screenshots. Five pre-trained architectures are tested, with MobileNetV1 achieving the best balance of accuracy (97.95% F1-score) and CPU speed (0.47s). For GPU-dependent, time-critical tasks, VGG16 is faster (0.04s) but less accurate (87.67%).},

keywords = {Image classification, Industrial Control System, Transfer Learning},

pubstate = {published},

tppubtype = {article}

}