Xabier Echeberria-Barrio
Research Subjects
Xabier Echeberria-Barrio has a total of 21 co-authors. Here are the top co-authors by number of shared publications:
- Francesco Zola
Digital Security Vicomtech Foundation Donostia Spain
4 shared publications - Raul Orduna-Urrutia
Digital Security Vicomtech Foundation Donostia Spain
4 shared publications - Amaia Gil
Fundacion Vicomtech, Basque Research and Technology Alliance (BRTA)
3 shared publications - Iñigo Mendialdua
Department of Computer Languages and Systems University of the Basque Country (UPV/EHU) Donostia-San Sebastian 20018 Spain
3 shared publications Raul Orduna-Urrutia
3 shared publicationsAmaia Gil-Lerchundi
2 shared publications
Latest Publications
- 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
- 18 September 2024
- Published by Association for Computing Machinery (ACM)
- p. 138-147
KINAITICS: Enhancing Cybersecurity Education Using AI-Based Tools and Gamification
- 18 September 2024
- Published by Association for Computing Machinery (ACM)
- p. 138-147
Open AccessKINAITICS: Enhancing Cybersecurity Education Using AI-Based Tools and Gamification
- Open Access
Topological safeguard for evasion attack interpreting the neural networks' behavior
In the last years, Deep Learning technology has been proposed in different fields, bringing many advances in each of them, but identifying new threats in these solutions regarding cybersecurity. Those implemented models have brought several vulnerabilities associated with Deep Learning technology. Moreover, those allow taking advantage of the implemented model, obtaining private information, and even modifying the model's decision-making. Therefore, interest in studying those vulnerabilities/attacks and designing defenses to avoid or fight them is gaining prominence among researchers. In particular, the widely known evasion attack is being analyzed by researchers; thus, several defenses to avoid such a threat can be found in the literature. Since the presentation of the L-BFG algorithm, this threat concerns the research community. However, it continues developing new and ingenious countermeasures since there is no perfect defense for all the known evasion algorithms. In this work, a novel detector of evasion attacks is developed. It focuses on the information of the activations of the neurons given by the model when an input sample is injected. Moreover, it puts attention to the topology of the targeted deep learning model to analyze the activations according to which neurons are connecting. This approach has been decided because the literature shows that the targeted model's topology contains essential information about if the evasion attack occurs. For this purpose, a huge data preprocessing is required to introduce all this information in the detector, which uses the Graph Convolutional Neural Network (GCN) technology. Thus, it understands the topology of the target model, obtaining promising results and improving the outcomes presented in the literature related to similar defenses.
Open AccessTopological safeguard for evasion attack interpreting the neural networks' behavior
In the last years, Deep Learning technology has been proposed in different fields, bringing many advances in each of them, but identifying new threats in these solutions regarding cybersecurity. Those implemented models have brought several vulnerabilities associated with Deep Learning technology. Moreover, those allow taking advantage of the implemented model, obtaining private information, and even modifying the model's decision-making. Therefore, interest in studying those vulnerabilities/attacks and designing defenses to avoid or fight them is gaining prominence among researchers. In particular, the widely known evasion attack is being analyzed by researchers; thus, several defenses to avoid such a threat can be found in the literature. Since the presentation of the L-BFG algorithm, this threat concerns the research community. However, it continues developing new and ingenious countermeasures since there is no perfect defense for all the known evasion algorithms. In this work, a novel detector of evasion attacks is developed. It focuses on the information of the activations of the neurons given by the model when an input sample is injected. Moreover, it puts attention to the topology of the targeted deep learning model to analyze the activations according to which neurons are connecting. This approach has been decided because the literature shows that the targeted model's topology contains essential information about if the evasion attack occurs. For this purpose, a huge data preprocessing is required to introduce all this information in the detector, which uses the Graph Convolutional Neural Network (GCN) technology. Thus, it understands the topology of the target model, obtaining promising results and improving the outcomes presented in the literature related to similar defenses.
- 20 November 2023
- Published by Elsevier in Pattern Recognition
- Vol. 147
Topological safeguard for evasion attack interpreting the neural networks’ behavior
- 20 November 2023
- Published by Elsevier in Pattern Recognition
- Vol. 147
Topological safeguard for evasion attack interpreting the neural networks’ behavior
- Open Access
- 10 March 2023
- Published by MDPI AG in Smart Cities
- Vol. 6 (2) , 897-912
ChronoEOS 2.0: Device Fingerprinting and EOSIO Blockchain Technology for On-Running Forensic Analysis in an IoT Environment
In industrial environments there are critical devices, so their correct operation must be ensured. In particular, having a secure record of the different events related to these devices is essential. Thus, this record can be used in future forensic investigations in case of accidents or production failures. In this sense, blockchain technology can bring reliability to the event log. In this paper, ChronoEOS 2.0, an extension of ChronoEOS, is presented. This new version can record the events that occur in multiple industrial robotic arms by deploying a Smart Contract in the EOSIO blockchain so that all events are immutably recorded in the blockchain. Furthermore, the new version allows using a unique fingerprint of the robot before registering an event in the blockchain. This fingerprint depends only on the characteristics of the operation and configuration of the robot. For this reason, ChronoEOS 2.0 not only increase the ability of ChronoEOS in terms of handling multiple devices but also increases the security and reliability of the operations. Finally, in this study, we verify that the new improvements have little impact on the hosting resources (RAM and Network are not altered, while CPU consumption is slightly higher due to the device fingerprinting module).
- 10 March 2023
- Published by MDPI AG in Smart Cities
- Vol. 6 (2) , 897-912
Open AccessChronoEOS 2.0: Device Fingerprinting and EOSIO Blockchain Technology for On-Running Forensic Analysis in an IoT Environment
In industrial environments there are critical devices, so their correct operation must be ensured. In particular, having a secure record of the different events related to these devices is essential. Thus, this record can be used in future forensic investigations in case of accidents or production failures. In this sense, blockchain technology can bring reliability to the event log. In this paper, ChronoEOS 2.0, an extension of ChronoEOS, is presented. This new version can record the events that occur in multiple industrial robotic arms by deploying a Smart Contract in the EOSIO blockchain so that all events are immutably recorded in the blockchain. Furthermore, the new version allows using a unique fingerprint of the robot before registering an event in the blockchain. This fingerprint depends only on the characteristics of the operation and configuration of the robot. For this reason, ChronoEOS 2.0 not only increase the ability of ChronoEOS in terms of handling multiple devices but also increases the security and reliability of the operations. Finally, in this study, we verify that the new improvements have little impact on the hosting resources (RAM and Network are not altered, while CPU consumption is slightly higher due to the device fingerprinting module).
- Open Access
- 7 April 2022
- Published by MDPI AG in Applied Sciences
- Vol. 12 (8) , 3725
Optimized Parameter Search Approach for Weight Modification Attack Targeting Deep Learning Models
Deep neural network models have been developed in different fields, bringing many advances in several tasks. However, they have also started to be incorporated into tasks with critical risks. That worries researchers who have been interested in studying possible attacks on these models, discovering a long list of threats from which every model should be defended. The weight modification attack is presented and discussed among researchers, who have presented several versions and analyses about such a threat. It focuses on detecting multiple vulnerable weights to modify, misclassifying the desired input data. Therefore, analysis of the different approaches to this attack helps understand how to defend against such a vulnerability. This work presents a new version of the weight modification attack. Our approach is based on three processes: input data clusterization, weight selection, and modification of the weights. Data clusterization allows a directed attack to a selected class. Weight selection uses the gradient given by the input data to identify the most-vulnerable parameters. The modifications are incorporated in each step via limited noise. Finally, this paper shows how this new version of fault injection attack is capable of misclassifying the desired cluster completely, converting the 100% accuracy of the targeted cluster to 0–2.7% accuracy, while the rest of the data continues being well-classified. Therefore, it demonstrates that this attack is a real threat to neural networks.
- 7 April 2022
- Published by MDPI AG in Applied Sciences
- Vol. 12 (8) , 3725
Open AccessOptimized Parameter Search Approach for Weight Modification Attack Targeting Deep Learning Models
Deep neural network models have been developed in different fields, bringing many advances in several tasks. However, they have also started to be incorporated into tasks with critical risks. That worries researchers who have been interested in studying possible attacks on these models, discovering a long list of threats from which every model should be defended. The weight modification attack is presented and discussed among researchers, who have presented several versions and analyses about such a threat. It focuses on detecting multiple vulnerable weights to modify, misclassifying the desired input data. Therefore, analysis of the different approaches to this attack helps understand how to defend against such a vulnerability. This work presents a new version of the weight modification attack. Our approach is based on three processes: input data clusterization, weight selection, and modification of the weights. Data clusterization allows a directed attack to a selected class. Weight selection uses the gradient given by the input data to identify the most-vulnerable parameters. The modifications are incorporated in each step via limited noise. Finally, this paper shows how this new version of fault injection attack is capable of misclassifying the desired cluster completely, converting the 100% accuracy of the targeted cluster to 0–2.7% accuracy, while the rest of the data continues being well-classified. Therefore, it demonstrates that this attack is a real threat to neural networks.
- 11 March 2022
- Published by Springer Nature in Neural Computing and Applications
- Vol. 34 (23) , 20477-20490
Understanding deep learning defenses against adversarial examples through visualizations for dynamic risk assessment
- 11 March 2022
- Published by Springer Nature in Neural Computing and Applications
- Vol. 34 (23) , 20477-20490
Understanding deep learning defenses against adversarial examples through visualizations for dynamic risk assessment
- Open Access
- 1 January 2021
- Published by Universidad Castilla la Mancha
Automating Intrusion Detection Systems in Smart Contracts
- 1 January 2021
- Published by Universidad Castilla la Mancha
Open AccessAutomating Intrusion Detection Systems in Smart Contracts
- 28 August 2020
- Published by Springer Nature
- p. 316-326
Deep Learning Defenses Against Adversarial Examples for Dynamic Risk Assessment
- 28 August 2020
- Published by Springer Nature
- p. 316-326
Deep Learning Defenses Against Adversarial Examples for Dynamic Risk Assessment
- Open Access
Generative Adversarial Networks for Bitcoin Data Augmentation
In Bitcoin entity classification, results are strongly conditioned by the ground-truth dataset, especially when applying supervised machine learning approaches. However, these ground-truth datasets are frequently affected by significant class imbalance as generally they contain much more information regarding legal services (Exchange, Gambling), than regarding services that may be related to illicit activities (Mixer, Service). Class imbalance increases the complexity of applying machine learning techniques and reduces the quality of classification results, especially for underrepresented, but critical classes. In this paper, we propose to address this problem by using Generative Adversarial Networks (GANs) for Bitcoin data augmentation as GANs recently have shown promising results in the domain of image classification. However, there is no "one-fits-all" GAN solution that works for every scenario. In fact, setting GAN training parameters is non-trivial and heavily affects the quality of the generated synthetic data. We therefore evaluate how GAN parameters such as the optimization function, the size of the dataset and the chosen batch size affect GAN implementation for one underrepresented entity class (Mining Pool) and demonstrate how a "good" GAN configuration can be obtained that achieves high similarity between synthetically generated and real Bitcoin address data. To the best of our knowledge, this is the first study presenting GANs as a valid tool for generating synthetic address data for data augmentation in Bitcoin entity classification.
Open AccessGenerative Adversarial Networks for Bitcoin Data Augmentation
In Bitcoin entity classification, results are strongly conditioned by the ground-truth dataset, especially when applying supervised machine learning approaches. However, these ground-truth datasets are frequently affected by significant class imbalance as generally they contain much more information regarding legal services (Exchange, Gambling), than regarding services that may be related to illicit activities (Mixer, Service). Class imbalance increases the complexity of applying machine learning techniques and reduces the quality of classification results, especially for underrepresented, but critical classes. In this paper, we propose to address this problem by using Generative Adversarial Networks (GANs) for Bitcoin data augmentation as GANs recently have shown promising results in the domain of image classification. However, there is no "one-fits-all" GAN solution that works for every scenario. In fact, setting GAN training parameters is non-trivial and heavily affects the quality of the generated synthetic data. We therefore evaluate how GAN parameters such as the optimization function, the size of the dataset and the chosen batch size affect GAN implementation for one underrepresented entity class (Mining Pool) and demonstrate how a "good" GAN configuration can be obtained that achieves high similarity between synthetically generated and real Bitcoin address data. To the best of our knowledge, this is the first study presenting GANs as a valid tool for generating synthetic address data for data augmentation in Bitcoin entity classification.