Cybersecurity Postgraduate Summer School

Professor Lorenzo Cavallaro – UCL

Title: Trustworthy AI... for Systems Security

Abstract: No day goes by without reading machine learning (ML) success stories across various application areas. Systems security is no exception, where ML’s tantalizing performance leave one to wonder whether there are any unsolved problems left. However, machine learning has no clairvoyant abilities and once the magic wears off, we’re left in uncharted territory. Is machine learning truly capable of ensuring systems security?

In this lecture, we will first illustrate some of the challenges in the context of adversarial ML evasion attacks against malware classifiers. We’ll see that the classic formulation is ill-suited for reasoning about how to generate realizable evasive malware. Then, we’ll provide a deep dive into recent work that provides a reformulation of the problem and enables more principled attack designs and defenses.

Implications are interesting, as the framework facilitates reasoning around end-to-end attacks that can generate real-world adversarial malware, at scale, that evades both vanilla and hardened classifiers, thus calling for novel defenses. As there is a deep connection between adversarial evasive attacks and out-of-distribution examples, we'll then broaden up our conversation highlighting the importance of reasoning beyond mere in-distribution performance by examining the consequences of distribution shifts in realistic settings. When relevant, we will also delve into behind-the-scenes aspects to encourage reflection on the reproducibility crisis. Our goal is to foster a deeper understanding of machine learning’s role in systems security and its potential for future advancements.

Bio: Lorenzo Cavallaro is a Full Professor of Computer Science at University College London (UCL), where he leads the [Systems Security Research Lab](https://s2lab.cs.ucl.ac.uk/). He grew up on pizza, spaghetti, and Phrack, and soon developed a passion for underground and academic research. Lorenzo’s research vision is to enhance the effectiveness of machine learning for systems security in adversarial settings. He works with his team to investigate the interplay among program analysis abstractions, representations, and ML models, and their crucial role in creating Trustworthy AI for Systems Security. Lorenzo publishes at and sits on the Program Committee of top-tier conferences in computer security and ML, received the Distinguished Paper Award at USENIX Security 2022, and an ICML 2024 Spotlight. He is also Associate Editor of ACM TOPS, IEEE TDSC, and Computer & Security. Lorenzo holds a PhD in Computer Science from the University of Milan and held positions at King’s College London, Royal Holloway University of London, Vrije Universitat Amsterdam, UC Santa Barbara, and Stony Brook University. In addition to his love for food, Lorenzo finds his Flow in science, music, and family.

Professor Juan Caballero – IMDEA Software

Lecture 1

Title: Malware & Cybercrime: An introduction

Abstract: Malicious software, or malware, is a key component of the cybercrime ecosystem leveraged by attackers to establish a permanent presence in compromised systems. This lecture provides an introduction to the use of malware in cybercrime, presents common malware classes, describes malware distribution techniques, and introduces malware classification approaches.

Lecture 2

Title: Malware & PUP: Keep them separate, if you can

Abstract: Potentially unwanted programs (PUP) such as adware and rogueware, while not outright malicious, exhibit intrusive behavior that generates user complaints and makes security vendors flag them as undesirable. PUP has been relatively little studied compared to malware despite recent indications that its prevalence may have surpassed that of malware. In this lesson, we present characteristics of PUP, measurements on PUP prevalence, an analysis of PUP distribution through pay-per-install (PPI) services, and the economics of PPI services distributing PUP.

Bio: Juan Caballero joined IMDEA Software as an Assistant Professor in November 2010, after obtaining his PhD in Electrical and Computer Engineering from Carnegie Mellon University, USA. Juan was a visiting postdoctoral researcher at the University of California, Berkeley for two years, starting in 2008. He was awarded the La Caixa fellowship for graduate studies in 2003. Juan's research focuses on cybersecurity, including security issues in systems, software, and networks. He enjoys designing program analysis techniques and applying them to analyze malicious code and build defenses against it. Additionally, he is interested in applying machine learning to security, network security, the economic aspects of cybercrime, and software engineering.

Professor Patricia Arias Cabarcos – Universität Paderborn

Title: Designing Security and Privacy for Everyone

Abstract: Living a secure digital life is not easy. Users need to deal with hundreds of passwords, ubiquitous cookie consent notices, and verbose privacy policies, among other complex security settings, to get their tasks done. This low usability of cybersecurity technologies is a well-known barrier to adoption, which translates into users being frequently hacked, phished, or victims of targeted attacks and privacy abuses. In this talk I will cover problems in achieving usable security and privacy for everybody, present state-of-the art research, and discuss open challenges.

Bio: Patricia Arias-Cabarcos is a Professor of IT Security at Paderborn University (UPB) and Principal Investigator at KASTEL. Her research interests focus on human-centered security and privacy, with a current emphasis on usable authentication, novel biometrics, behavioral data protection, and data-driven transparency. The vision guiding her research group is that people should not experience cognitive stress or need deep technical knowledge to live a secure digital life. Before joining UPB, she was a senior researcher at the Karlsruhe Institute of Technology (2019-2021), a Humboldt Fellow at the University of Mannheim (2017-2019), and an Assistant Professor (2013-2018) at Carlos III University of Madrid (Spain). She has published over 60 papers in top international venues, including Usenix Security and ACM CCS.

Assistant Professor Maura Pintor – University of Cagliari

Title: Reliable Evaluation and Benchmarking of Machine Learning Models

Abstract: Rigorous evaluation of machine learning (ML) models is essential before deployment. To understand ML's sensitivity to adversarial attacks and counter their effects, ML model designers craft worst-case perturbations and test them against their products. However, many of the proposed defenses have been shown to provide a false sense of security due to failures of the attacks rather than actual robustness. To this end, it's important to set up trustworthy evaluation tools. In this talk, we will investigate existing benchmarking tools and we will highlight their issues, avoiding known mistakes to ensure high-quality evaluations. Moreover, current ML benchmarks are a first step, but they only offer an in-vitro evaluation. Addressing practical aspects like how predictions react to data drift over time and model updates is also important in real-world applications. For this reason, we will provide insights on analyzing how both performance and robustness evolve over time. Finally, we will discuss new testing and benchmarking guidelines to develop novel techniques to ensure models behave robustly in real-world scenarios, where not only are they the target of attacks, but they are also subject to data drifts and situations unseen in training.

Bio: Maura Pintor is an Assistant Professor at the PRA Lab, in the Department of Electrical and Electronic Engineering of the University of Cagliari, Italy. She received her PhD in Electronic and Computer Engineering from the University of Cagliari in 2022. Her main research interests revolve around providing trustworthy security evaluations of ML. She was a visiting student at Eberhard Karls Universitaet Tuebingen, Germany, from March to June 2020 and at the Software Competence Center Hagenberg (SCCH), Austria, from May to August 2021. She is currently visiting researcher at the Universitat Autònoma de Barcelona (UAB), at the Computer Vision Center (CVC), until the end of October 2024. She is reviewer for ACM CCS, ICLR, NeurIPS, ACSAC, ICCV, and for several Q1 journals. She is co-chair of the ACM Workshop on Artificial Intelligence and Security (AISec), co-located with ACM CCS.

Associate Professor Sergio Pastrana – UC3M

Title: The Art of Cybercrime Community Research

Abstract: Cybercrime is a growing problem with considerable impact on modern societies. One key factor is the proliferation of online cybercrime communities, where actors trade products and services, and also learn from peers. As such, understanding the operation and behaviour of these communities is of great interests for various stakeholders, including academics, LEAs and industry practitioners. In this talk, I will present some of the challenges and approaches taken to study these communities, including the activities being discussed, and the actors involved.

Bio: Sergio is Associate Professor at the Computer Security Lab (COSEC) from the University Carlos III of Madrid. Formerly. He worked in the Cambridge Cybercrime Centre from the University of Cambridge. His research interests are manyfold through the field of cybersecurity and cybercrime. He received a PhD in Computer Science and Technology in 2014 by University of Carlos III de Madrid, where he teach courses in official studies offered by the University Carlos III about Cybersecurity, as well as ocassional participation in Master courses, summer schools and workshops.

Dr Budi Arief – University of Kent

Title: A Socio-technical Look into Ransomware

Abstract: Ransomware is a form of malware designed to prevent its victim's access to their data or to leak victim's sensitive data, unless the victim pays the ransom demand to the attacker. Ransomware has been around for quite a while now, and it remains one of the most prevalent cyber threats to individuals and, of growing concern, businesses. On top of the the technical elements, there are many human aspects involved in ransomware operations. As such, it is important to understand the socio-technical perspectives involved, in order to devise more effective countermeasures.

This talk covers a wide range of socio-technical research we have conducted in combatting ransomware, ranging from investigating ransomware deployment methods to roadmapping potential countermeasures (and why some of them might fail). With the popularity of Internet of Things (IoT) devices, we have also explored the feasibility of ransomware attacks on IoT devices, and the potential consequences that such attacks might bring.

Finally, we have been looking to better understand the stakeholders involved, for instance by investigating how attackers might get started in ransomware operations and their justification for taking part in this crime, as well as by exploring various human factors that can contribute to victimisation (including ransom notes designs and victims' personality types). The talk will wrap up with a quick peek into our ongoing work focusing on the threat of ransomware in the Industrial IoT (IIoT) domain, which is part of the "Countering HArms caused by Ransomware in the Internet Of Things (CHARIOT)" project funded by the UK Engineering and Physical Sciences Research Council (EPSRC) and the Research Institute in Trustworthy Inter-Connected Cyber-Physical Systems (RITICS).

Bio: Dr. Budi Arief is a Senior Lecturer at the School of Computing and Operations Lead at the Institute of Cyber Security for Society (iCSS) at the University of Kent (UK). His research focuses on cybercrime and computer security, particularly ransomware, Internet of Things security, and combating child sexual abuse, with an interdisciplinary approach. He has been the principal investigator on EPSRC-funded projects like CHARIOT and co-investigator on EMPHASIS. Currently, he leads the EU-funded project HEROES and participates in ALUNA. He has published over 70 articles in prominent journals and conferences such as Computers & Security, IEEE Security & Privacy, ACM CCS, and IEEE TrustCom. Budi earned his BSc and PhD in Computing Science from Newcastle University (UK).

Assistant Professor Guillermo Suarez-Tangil – IMDEA Networks

Title: Interrogating Singularities: from cloud-based apps to LLMs

Abstract: In physics, a singularity refers to a point in space-time where the gravitational pull becomes infinitely strong and the curvature of space-time becomes infinite. This is similar to the center of a black hole where the density is infinitely high and conventional physical laws break down.

This talk will offer an overview of mechanism design to measure singularities in the cloud. This is, when you have to analyze a service, a program, or a component for which you do not have access to its internals and you need to use scarce interrogative methodologies to extract the security and privacy attributes from it.

Bio: Guillermo Suarez-Tangil is an Assistant Professor at IMDEA Networks and a Ramon y Cajal Fellow. His research focuses on modeling emerging threats in online communities and devising effective mitigation strategies. Before joining IMDEA, Guillermo was a Lecturer at King's College London (KCL). He has been a senior research associate at University College London (UCL), where he explored the use of program analysis to study malware. He has also been actively involved in other research directions aiming at detecting and preventing Mass-Marketing Fraud (MMF) and security and privacy on the social web. Prior to that, he held a post-doctoral position at Royal Holloway, University of London (RHUL), where he was involved in detecting malware and improving security in the mobile age (MobSec). He also holds solid expertise in building novel data-learning algorithms for malware analysis. He obtained his PhD in smart malware analysis at Carlos III University of Madrid with distinction and received the Best National Student Academic Award—a competitive award given to the best Thesis in the field of Engineering between 2014-2015 with about a 1% acceptance rate (about 100 Cum Laude Theses were invited to compete for the only award).