Lander Segurola-Gil
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Lander Segurola-Gil has a total of 17 co-authors. Here are the top co-authors by number of shared publications:
- Francesco Zola
Digital Security Vicomtech Foundation Donostia Spain
6 shared publications - Raul Orduna-Urrutia
Digital Security Vicomtech Foundation Donostia Spain
4 shared publications - Ane Miren Florez-Tapia
Digital Security, Vicomtech, Basque Research and Technology Alliance, Donostia-San Sebastián, Spain
3 shared publications - Itziar Irigoien
Department of Computer Science and Artificial Intelligence, University of the Basque Country (UPV/EHU), Donostia-San Sebastián, Spain
3 shared publications - Mikel Moreno-Moreno
Digital Security, Vicomtech, Basque Research and Technology Alliance, Donostia-San Sebastián, Spain
3 shared publications - Idoia Eizaguirre-Peral
Basque Research and Technology Alliance (BRTA) 20009 Vicomtech Donostia San Sebastian Vicomtech Spain
2 shared publications
Latest Publications
- 24 August 2026
- Published by Springer Nature in The Journal of Supercomputing
- Vol. 82 (13) , 662
Graph entropy-based node exposure score for risk assessment in information systems
- 24 August 2026
- Published by Springer Nature in The Journal of Supercomputing
- Vol. 82 (13) , 662
Graph entropy-based node exposure score for risk assessment in information systems
- Open Access
- 27 February 2026
- Published by Elsevier in SSRN Electronic Journal
Topological Interpretability in Convolution Neural Network Behavior for Adversarial Evasion Protection
Convolutional neural networks (CNNs) have achieved outstanding performance in domains such as medical imaging and autonomous systems. However, their limited interpretability continues to pose security concerns, particularly under adversarial manipulation. This work extends the application of interpretability-based defense mechanisms by introducing a convolution-focused variant that shifts the analysis from the classifier to the feature extraction layers. Specifically, it represents the behavior of convolutional layers through a graph-based model, where topological descriptors characterize internal CNN dynamics. Using filter-level attributes derived from previous neuron-level interpretability work, the method enables the detection of anomalies in feature extraction patterns caused by adversarial inputs. The approach is evaluated on a VGG16 network trained on a breast cancer histopathology dataset and a CNN trained on MNIST. Results show that the convolution-focused detector achieves performance comparable to, and in some scenarios surpassing, classifier-based detectors, providing valuable insight into the layers most affected by attacks. Moreover, the method can be applied to CNN architectures with pretrained extractors, ensuring robustness without introducing weaknesses during fine-tuning.
- 27 February 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessTopological Interpretability in Convolution Neural Network Behavior for Adversarial Evasion Protection
Convolutional neural networks (CNNs) have achieved outstanding performance in domains such as medical imaging and autonomous systems. However, their limited interpretability continues to pose security concerns, particularly under adversarial manipulation. This work extends the application of interpretability-based defense mechanisms by introducing a convolution-focused variant that shifts the analysis from the classifier to the feature extraction layers. Specifically, it represents the behavior of convolutional layers through a graph-based model, where topological descriptors characterize internal CNN dynamics. Using filter-level attributes derived from previous neuron-level interpretability work, the method enables the detection of anomalies in feature extraction patterns caused by adversarial inputs. The approach is evaluated on a VGG16 network trained on a breast cancer histopathology dataset and a CNN trained on MNIST. Results show that the convolution-focused detector achieves performance comparable to, and in some scenarios surpassing, classifier-based detectors, providing valuable insight into the layers most affected by attacks. Moreover, the method can be applied to CNN architectures with pretrained extractors, ensuring robustness without introducing weaknesses during fine-tuning.
- Open Access
- 7 March 2025
- Published by Springer Nature in International Journal of Machine Learning and Cybernetics
- Vol. 16 (11) , 9731
Correction: Unsupervised Anomaly Detection Approach for Cyberattack Identification
- 7 March 2025
- Published by Springer Nature in International Journal of Machine Learning and Cybernetics
- Vol. 16 (11) , 9731
Open AccessCorrection: Unsupervised Anomaly Detection Approach for Cyberattack Identification
- 16 November 2024
- Published by Springer Nature
- p. 88-97
Generation of Industrial Protocol-Traffic via Enhanced Wasserstein GAN
- 16 November 2024
- Published by Springer Nature
- p. 88-97
Generation of Industrial Protocol-Traffic via Enhanced Wasserstein GAN
- Open Access
- 17 June 2024
- Published by Springer Nature in International Journal of Machine Learning and Cybernetics
- Vol. 15 (11) , 5291-5302
Unsupervised Anomaly Detection Approach for Cyberattack Identification
- 17 June 2024
- Published by Springer Nature in International Journal of Machine Learning and Cybernetics
- Vol. 15 (11) , 5291-5302
Open AccessUnsupervised Anomaly Detection Approach for Cyberattack Identification
- Open Access
Conditional Generative Adversarial Network for keystroke presentation attack
Cybersecurity is a crucial step in data protection to ensure user security and personal data privacy. In this sense, many companies have started to control and restrict access to their data using authentication systems. However, these traditional authentication methods, are not enough for ensuring data protection, and for this reason, behavioral biometrics have gained importance. Despite their promising results and the wide range of applications, biometric systems have shown to be vulnerable to malicious attacks, such as Presentation Attacks. For this reason, in this work, we propose to study a new approach aiming to deploy a presentation attack towards a keystroke authentication system. Our idea is to use Conditional Generative Adversarial Networks (cGAN) for generating synthetic keystroke data that can be used for impersonating an authorized user. These synthetic data are generated following two different real use cases, one in which the order of the typed words is known (ordered dynamic) and the other in which this order is unknown (no-ordered dynamic). Finally, both keystroke dynamics (ordered and no-ordered) are validated using an external keystroke authentication system. Results indicate that the cGAN can effectively generate keystroke dynamics patterns that can be used for deceiving keystroke authentication systems.
Open AccessConditional Generative Adversarial Network for keystroke presentation attack
Cybersecurity is a crucial step in data protection to ensure user security and personal data privacy. In this sense, many companies have started to control and restrict access to their data using authentication systems. However, these traditional authentication methods, are not enough for ensuring data protection, and for this reason, behavioral biometrics have gained importance. Despite their promising results and the wide range of applications, biometric systems have shown to be vulnerable to malicious attacks, such as Presentation Attacks. For this reason, in this work, we propose to study a new approach aiming to deploy a presentation attack towards a keystroke authentication system. Our idea is to use Conditional Generative Adversarial Networks (cGAN) for generating synthetic keystroke data that can be used for impersonating an authorized user. These synthetic data are generated following two different real use cases, one in which the order of the typed words is known (ordered dynamic) and the other in which this order is unknown (no-ordered dynamic). Finally, both keystroke dynamics (ordered and no-ordered) are validated using an external keystroke authentication system. Results indicate that the cGAN can effectively generate keystroke dynamics patterns that can be used for deceiving keystroke authentication systems.
- Open Access
- 27 October 2022
- Published by Association for Computing Machinery (ACM)
- p. 119-126
Keystroke Presentation Attack: Generative Adversarial Networks for replacing user behaviour
Digital security has become crucial in this new era of technology and biometry is becoming a natural and reliable authentication system. In recent years, keystroke dynamics, a type of behavioral biometric, has been used for user authentication and attack detection. In this study, we pursue a new approach to keystroke dynamics data generation focused on the impersonation of a user at the identification stage using Conditional Generative Adversarial Networks (cGAN). To that aim, three different architectures have been designed, implemented, and validated: a Vanilla-cGAN based on simple Neural Networks (NN), an LSTM-cGAN based on Recurrent Neural Networks using Long Short-Term Memory units (LSTM), and a CNN-cGAN based on Convolutional Neural Networks. These models have been validated in two different conditions, one in which the attacker knows exactly the order of the typed words for replicating the behavior and the other in which the order is unknown. To validate the data generated by these models, beyond the internal discriminator’s accuracy, a pre-trained Siamese Network has been used to detect whether two keystroke sequences belong to the same or not. This study suggests that the keystroke dynamics of a user can be successfully imitated via keystroke dynamics data generation using cGANs with different architectures.
- 27 October 2022
- Published by Association for Computing Machinery (ACM)
- p. 119-126
Open AccessKeystroke Presentation Attack: Generative Adversarial Networks for replacing user behaviour
Digital security has become crucial in this new era of technology and biometry is becoming a natural and reliable authentication system. In recent years, keystroke dynamics, a type of behavioral biometric, has been used for user authentication and attack detection. In this study, we pursue a new approach to keystroke dynamics data generation focused on the impersonation of a user at the identification stage using Conditional Generative Adversarial Networks (cGAN). To that aim, three different architectures have been designed, implemented, and validated: a Vanilla-cGAN based on simple Neural Networks (NN), an LSTM-cGAN based on Recurrent Neural Networks using Long Short-Term Memory units (LSTM), and a CNN-cGAN based on Convolutional Neural Networks. These models have been validated in two different conditions, one in which the attacker knows exactly the order of the typed words for replicating the behavior and the other in which the order is unknown. To validate the data generated by these models, beyond the internal discriminator’s accuracy, a pre-trained Siamese Network has been used to detect whether two keystroke sequences belong to the same or not. This study suggests that the keystroke dynamics of a user can be successfully imitated via keystroke dynamics data generation using cGANs with different architectures.
- Open Access
- 8 June 2022
- Published by Academic Conferences and Publishing International Ltd in European Conference on Cyber Warfare and Security
- Vol. 21 (1) , 271-279
Siamese Neural Network and Machine Learning for DGA Classification
Domain Generation Algorithms (DGA) are systems used to create immediate multiple and varying domain names. Such “artificial” domains can be then used for siting command and control servers which in turn oversee recruiting/infecting devices, and finally turning them into new resources to be exploited. In this sense, identifying DGA domain names can be crucial, to avoid cyberattacks like Phishing, Spam sending, Bitcoin mining, and many other. Usually, domain names generated by DGAs, are comprised by illegible character strings, but new “intelligent” DGAs tend to generate names using combination of words in dictionaries making its detection a challenging task. For this reason, in this work, we propose to address this problem using a combination of Machine Learning algorithms for improving the classification of DGAs domains. In particular, we propose to combine Siamese Neural Networks and traditional supervised Machine Learning algorithms in order to expand the input domain into separable n-dimensional data points and then achieve the domain classification. The proposed approach can be separated into 3 phases. In a first phase, domain names are encoded, by a one-hot encoder and a variation of this, named probabilistic one-hot encoder, which are implemented separately. Then, in the second phase, Long Short-Term Memory and Convolutional Siamese embedders are tested and compared. In particular, the first one is combined with the one-hot, while the Convolution algorithm is applied with the probabilistic one-hot encoded data. In the final step, five Machine Learning algorithms are tested using the two ways embedded data. Both embedder approaches reach very high results in terms of F1-score and Accuracy (about 91%) depending on the implemented classifier. The promising results obtained by the application of the proposed method shows that it is possible to perform DGA domain classification uniquely over the domain names, without considering external information such as DNS packets features.
- 8 June 2022
- Published by Academic Conferences and Publishing International Ltd in European Conference on Cyber Warfare and Security
- Vol. 21 (1) , 271-279
Open AccessSiamese Neural Network and Machine Learning for DGA Classification
Domain Generation Algorithms (DGA) are systems used to create immediate multiple and varying domain names. Such “artificial” domains can be then used for siting command and control servers which in turn oversee recruiting/infecting devices, and finally turning them into new resources to be exploited. In this sense, identifying DGA domain names can be crucial, to avoid cyberattacks like Phishing, Spam sending, Bitcoin mining, and many other. Usually, domain names generated by DGAs, are comprised by illegible character strings, but new “intelligent” DGAs tend to generate names using combination of words in dictionaries making its detection a challenging task. For this reason, in this work, we propose to address this problem using a combination of Machine Learning algorithms for improving the classification of DGAs domains. In particular, we propose to combine Siamese Neural Networks and traditional supervised Machine Learning algorithms in order to expand the input domain into separable n-dimensional data points and then achieve the domain classification. The proposed approach can be separated into 3 phases. In a first phase, domain names are encoded, by a one-hot encoder and a variation of this, named probabilistic one-hot encoder, which are implemented separately. Then, in the second phase, Long Short-Term Memory and Convolutional Siamese embedders are tested and compared. In particular, the first one is combined with the one-hot, while the Convolution algorithm is applied with the probabilistic one-hot encoded data. In the final step, five Machine Learning algorithms are tested using the two ways embedded data. Both embedder approaches reach very high results in terms of F1-score and Accuracy (about 91%) depending on the implemented classifier. The promising results obtained by the application of the proposed method shows that it is possible to perform DGA domain classification uniquely over the domain names, without considering external information such as DNS packets features.
- Open Access
- 1 April 2022
- Published by Springer Nature in Applied Intelligence
- Vol. 52 (15) , 17289-17314
Attacking Bitcoin anonymity: generative adversarial networks for improving Bitcoin entity classification
Classification of Bitcoin entities is an important task to help Law Enforcement Agencies reduce anonymity in the Bitcoin blockchain network and to detect classes more tied to illegal activities. However, this task is strongly conditioned by a severe class imbalance in Bitcoin datasets. Existing approaches for addressing the class imbalance problem can be improved considering generative adversarial networks (GANs) that can boost data diversity. However, GANs are mainly applied in computer vision and natural language processing tasks, but not in Bitcoin entity behaviour classification where they may be useful for learning and generating synthetic behaviours. Therefore, in this work, we present a novel approach to address the class imbalance in Bitcoin entity classification by applying GANs. In particular, three GAN architectures were implemented and compared in order to find the most suitable architecture for generating Bitcoin entity behaviours. More specifically, GANs were used to address the Bitcoin imbalance problem by generating synthetic data of the less represented classes before training the final entity classifier. The results were used to evaluate the capabilities of the different GAN architectures in terms of training time, performance, repeatability, and computational costs. Finally, the results achieved by the proposed GAN-based resampling were compared with those obtained using five well-known data-level preprocessing techniques. Models trained with data resampled with our GAN-based approach achieved the highest accuracy improvements and were among the best in terms of precision, recall and f1-score. Together with Random Oversampling (ROS), GANs proved to be strong contenders in addressing Bitcoin class imbalance and consequently in reducing Bitcoin entity anonymity (overall and per-class classification performance). To the best of our knowledge, this is the first work to explore the advantages and limitations of GANs in generating specific Bitcoin data and “attacking” Bitcoin anonymity. The proposed methods ultimately demonstrate that in Bitcoin applications, GANs are indeed able to learn the data distribution and generate new samples starting from a very limited class representation, which leads to better detection of classes related to illegal activities.
- 1 April 2022
- Published by Springer Nature in Applied Intelligence
- Vol. 52 (15) , 17289-17314
Open AccessAttacking Bitcoin anonymity: generative adversarial networks for improving Bitcoin entity classification
Classification of Bitcoin entities is an important task to help Law Enforcement Agencies reduce anonymity in the Bitcoin blockchain network and to detect classes more tied to illegal activities. However, this task is strongly conditioned by a severe class imbalance in Bitcoin datasets. Existing approaches for addressing the class imbalance problem can be improved considering generative adversarial networks (GANs) that can boost data diversity. However, GANs are mainly applied in computer vision and natural language processing tasks, but not in Bitcoin entity behaviour classification where they may be useful for learning and generating synthetic behaviours. Therefore, in this work, we present a novel approach to address the class imbalance in Bitcoin entity classification by applying GANs. In particular, three GAN architectures were implemented and compared in order to find the most suitable architecture for generating Bitcoin entity behaviours. More specifically, GANs were used to address the Bitcoin imbalance problem by generating synthetic data of the less represented classes before training the final entity classifier. The results were used to evaluate the capabilities of the different GAN architectures in terms of training time, performance, repeatability, and computational costs. Finally, the results achieved by the proposed GAN-based resampling were compared with those obtained using five well-known data-level preprocessing techniques. Models trained with data resampled with our GAN-based approach achieved the highest accuracy improvements and were among the best in terms of precision, recall and f1-score. Together with Random Oversampling (ROS), GANs proved to be strong contenders in addressing Bitcoin class imbalance and consequently in reducing Bitcoin entity anonymity (overall and per-class classification performance). To the best of our knowledge, this is the first work to explore the advantages and limitations of GANs in generating specific Bitcoin data and “attacking” Bitcoin anonymity. The proposed methods ultimately demonstrate that in Bitcoin applications, GANs are indeed able to learn the data distribution and generate new samples starting from a very limited class representation, which leads to better detection of classes related to illegal activities.
- 1 January 2021
- Published by Springer Nature
- p. 55-70
NBcoded: Network Attack Classifiers Based on Encoder and Naive Bayes Model for Resource Limited Devices
- 1 January 2021
- Published by Springer Nature
- p. 55-70
NBcoded: Network Attack Classifiers Based on Encoder and Naive Bayes Model for Resource Limited Devices