Published in ArXiV, 2024
In this paper we investigate the impact of packing on the performance of machine learning-based malware detection and classification systems.
Recommended citation: Daniel Gibert, Nikolaos Totosis, Constantinos Patsakis, Giulio Zizzo, Quan Le. "Assessing the Impact of Packing on Machine Learning-Based Malware Detection and Classification Systems." ArXiV.
Download Paper
Published in IEEE Access, 2024
In this work, we propose a chunk-based smoothing scheme that reduces the chances of sampling adversarial content injected by malware authors by selecting correlated subsets of bytes. During training, our chunk-based smoothing scheme trains a base classifier to make classifications on a subset of contiguous bytes or chunk of bytes. At test time, a large number of chunks are then classified by a base classifier and the consensus among these classifications is then reported as the final prediction. We propose two strategies to determine the location of the chunks used for classification: (1) randomly selecting the locations of the chunks and (2) selecting contiguous adjacent chunks. To showcase the effectiveness of our approach, we have trained two classifiers with our chunk-based smoothing schemes on the BODMAS dataset. Our findings reveal that the chunk-based smoothing classifiers exhibit greater resilience against adversarial malware examples generated with state-of-the-art evasion attacks, outperforming a non-smoothed classifier and a randomized smoothing-based classifier by a great margin.
Recommended citation: Daniel Gibert, Giulio Zizzo, Quan Le, Jordi Planes. (2024). "Adversarial Robustness of Deep Learning-based Malware Detectors via (De) Randomized Smoothing." IEEE Access.
Download Paper
Published in Springer, 2024
Chapter that appears in the Book “Malware: Handbook of Prevention and Detection”, published by Springer. The chapter explores how machine learning has been applied to build malware detection systems for the Windows operating system. The chapter starts by introducing the main components of a Machine Learning pipeline, highlighting the challenges of collecting and maintaining up-to-date datasets. Following this introduction, various state-of-the-art malware detectors are presented, encompassing both feature-based and deep learning-based detectors. Subsequent sections introduce the primary challenges encountered by machine learning-based malware detectors, including concept drift and adversarial attacks. Lastly, this chapter concludes by providing a brief overview of the ongoing research on adversarial defenses.
Recommended citation: Gibert, D. (2025). "Machine Learning for Windows Malware Detection and Classification: Methods, Challenges and Ongoing Research." In: Gritzalis, D., Choo, KK.R., Patsakis, C. (eds) Malware. Advances in Information Security, vol 91. Springer, Cham. https://doi.org/10.1007/978-3-031-66245-4_6 .
Download Paper
Published in ArXiv, 2024
This paper extends our AISEC paper with a preprocessing step that allows us to provide robustness certificates against patch, append, and content injection attacks.
Recommended citation: Daniel Gibert, Luca Demetrio, Giulio Zizzo, Quan Le, Jordi Planes, Battista Biggio. (2024). "Certified Adversarial Robustness of Machine Learning-based Malware Detectors via (De) Randomized Smoothing." ArXiv
Download Paper
Published in Applied Thermal Engineering, 2024
In this paper we analyze the competitiveness of deep reinforcement learning to control a solar-driven seasonal sorption TES system in comparison to a traditional optimized rule-based control strategy.
Recommended citation: Alicia Crespo, Daniel Gibert, Álvaro de Gracia, Cèsar Fernández. "Optimal control of a solar-driven seasonal sorption storage system through deep reinforcement learning." Applied Thermal Engineering.
Download Paper
Published in Journal Expert Systems With Applications, 2022
In this work, we present an hybrid system for malware classification that combines feature engineering and deep learning using an early-fusion mechanism.
Recommended citation: Daniel Gibert, Jordi Planes, Carles Mateu, Quan Le. (2022). "Fusing feature engineering and deep learning: A case study for malware classification." Journal Expert Systems With Applications.
Download Paper
Published in Journal Computers & Security, 2022
This paper explores the vulnerability of classifiers against the dead code insertion technique and it poposes a reinforcement learning framework to bypass malware classifers
Recommended citation: Daniel Gibert, Matt Fredrikson, Carles Mateu, Jordi Planes, Quan Le. (2022). "Enhancing the insertion of NOP instructions to obfuscate malware via deep reinforcement learning." Journal Computers & Security.
Download Paper
Published in Journal Computers & Security, 2021
This paper presents an exhaustive evaluation of the state-of-the-art approaches for malware classification against common metamorphic attacks.
Recommended citation: Daniel Gibert, Carles Mateu, Jordi Planes, Joao Marques-Silva. (2021). "Auditing static machine learning anti-Malware tools against metamorphic attacks." Journal Computers & Security.
Download Paper
Published in Journal Computers & Security, 2020
We present HYDRA, a multimodal deep learning framework for malware classification that combines N-gram like features extracted from bytes and opcodes sequences and API-based features.
Recommended citation: Daniel Gibert, Carles Mateu, Jordi Planes. (2020). "HYDRA: A multimodal deep learning framework for malware classification." Journal Computers & Security.
Download Paper
Published in Journal of Network and Computer Applciations, 2020
In this paper, we present a systematic review of machine learning-based approaches for malware detection, classified into static, dynamic, and hybrid approaches.
Recommended citation: Daniel Gibert, Carles Mateu, Jordi Planes. (2020). "The rise of machine learning for detection and classification of malware: Research developments, trends and challenges." Journal of Network and Computer Applications.
Download Paper
Published in Journal of Computer Virology and Hacking Techniques, 2019
In this paper, we propose a CNN-based approach for malware classification based on the representation of malware as grayscale images.
Recommended citation: Daniel Gibert, Carles Mateu, Jordi Planes. (2019). "Using convolutional neural networks for classification of malware represented as images." Journal of Computer Virology and Hacking Techniques.
Download Paper
Published in ESORICS 2023 International Workshops. Workshops on Security and Artificial Intelligence (SECAI). Lecture Notes in Computer Science. Springer, 2024
This paper presents a defense against adversarial EXEmples based on randomized smoothing.
Recommended citation: Daniel Gibert, Giulio Zizzo and Quan Le. (2023). "Towards a Practical Defense Against Adversarial Attacks on Deep Learning-Based Malware Detectors via Randomized Smoothing." Workshop on Security and Artificial Intelligence (SECAI 2023).
Download Paper
Published in IEEE European Symposium on Security and Privacy Workshops (EuroS&PW), 2023
This work presents a query-free approach to craft adversarial malware examples to evade ML-based malware detectors using Generative Adversarial Networks.
Recommended citation: Daniel Gibert, Jordi Planes, Quan Le, Giulio Zizzo. (2023). "A Wolf in Sheep’s Clothing: Query-Free Evasion Attacks Against Machine Learning-Based Malware Detectors with Generative Adversarial Networks." 2023 IEEE European Symposium on Security and Privacy Workshops (EuroS&PW).
Download Paper
Published in 16th ACM Workshop on Artificial Intelligence and Security (AISEC), 2023
This paper presents an adversarial defense based on (de)randomized smoothing that provides deterministic robustness certificates against patch and append attacks on end-to-end malware detectors.
Recommended citation: Daniel Gibert, Giulio Zizzo and Quan Le. (2023). "Certified robustness of static deep learning-based malware detectors against patch and append attacks." 16th ACM Workshop on Artificial Intelligence and Security (AISEC 2023).
Download Paper
Published in International Joint Conference on Neural Networks (IJCNN), 2020
We introduce Orthrus, a bimodal approach for malware categorization based on deep learning..
Recommended citation: Daniel Gibert, Carles Mateu, Jordi Planes. (2020). "Orthrus: A Bimodal Learning Architecture for Malware Classification." IJCNN 2020.
Download Paper
Published in IAAI-AAAI, 2018
We introduce a deep learning approach for malware classification based on the structural entropy representation of Portable Executable files.
Recommended citation: Daniel Gibert, Carles Mateu, Jordi Planes. (2018). "Classification of malware by using structural entropy on convolutional neural networks." IAAI-AAAI 2018.
Download Paper
Published in CCIA, 2017
We propose a CNN architecture that learns N-gram like features from malware assembly language source code.
Recommended citation: Daniel Gibert, Carles Mateu, Jordi Planes, Daniel Solis, Ramon Vicens (2017). "Convolutional neural networks for classification of malware assembly code." CCIA 2017.
Download Paper