June 29, 2020
15:00 – 18:30 (CEST).
The DSN Workshop on Dependable and Secure Machine Learning (DSML) is an open forum for researchers, practitioners, and regulatory experts, to present and discuss innovative ideas and practical techniques and tools for producing dependable and secure machine learning (ML) systems. A major goal of the workshop is to draw the attention of the research community to the problem of establishing guarantees of reliability, security, safety, and robustness for systems that incorporate increasingly complex ML models, and to the challenge of determining whether such systems can comply with the requirements for safety-critical systems. A further goal is to build a research community at the intersection of machine learning and dependable and secure computing.
The program of the workshop is available here.
Video of the session with Q&A
Welcome and Keynote 1
Dependability and security are of the utmost importance for computing systems. Due to the scale and complexity of current systems, both aspects are a permanent and growing concern in industry and academia. On the one hand, the volume and diversity of functional and non-functional data, including open source information, along with increasingly dynamical operating environments, create additional obstacles to the dependability and security of systems. On the other hand, it creates an information rich environment that, leveraged by techniques from modern data science, machine and statistical learning, and visualization, will contribute to improve systems resilience in contexts of dynamic operating environments and unexpected operating conditions. As such, there is a strong demand for production-ready systems leveraging from data-centric solutions able to improve and, adaptively, maintain the dependability and security of computing systems.
The workshop on Data-Centric Dependability and Security (DCDS) aims at providing researchers with a forum to exchange and discuss scientific contributions and open challenges, both theoretical and practical, related to the use of data-centric approaches that promote the dependability and cybersecurity of computing systems. We want to foster joint work and knowledge exchange between the dependability and security communities, and researchers and practitioners from areas such as machine and statistical learning, and data science and visualization. The workshop provides a forum for discussing novel trends in data-centric processing technologies and the role of such technologies in the development of resilient systems. It aims to discuss novel approaches for processing and analysing data generated by the systems as well as information gathered from open sources, leveraging from data science, machine and statistical learning techniques, and visualization. The workshop shall contribute to identify new application areas as well as open and future research problems, for data-centric approaches to system dependability and security.
The program of the workshop can be checked here.
Videos of the workshop sessions including Q&A
Welcome DCDS 2020 and Session 1: Keynote
Session 2: Network Security & Privacy
List of Papers in the session and their Teasers
A number of high-performance computing (HPC) commercial off-the-shelf (COTS) platforms offer the computation capabilities needed by autonomous systems in domains such as automotive, space, avionics, robotics and factory automation by means of multicores, GPUs and specialized accelerators. Unfortunately, the utilization of HPC platforms has been traditionally considered out of the reach for the safety-critical systems industry due to the difficulties or roadblocks these platforms bring to the certification process. This workshop focuses on the research towards the adoption of HPC hardware and software platforms in the context of safety- and security-critical applications. In particular, the scope of the workshop includes functional-safety and security requirements for HPC systems, including but not limited to non-functional aspects such as time predictability and energy consumption.
Topics of interest
- High-performance critical real-time systems
- Dependable systems and safety mechanisms
- Hardware and Software security in safety-critical systems
Carles Hernàndez (Universitat Politènica de València)
Jaume Abella (Barcelona Supercomputing Center)
Mikel Azkarate-askatsua (Ikerlan)
Roman Obermaisser (University of Siegen)
The program of the workshop can be checked here.
Video containing all workshop sessions
Over the last years, aerial and ground vehicles as well as mobile robot systems have been receiving an increased number of electronic components, connected through wireless networks and running embedded software. As processing power increases and software becomes more sophisticated, these vehicles gain the ability to perform complex operations, becoming more autonomous, safe, efficient, adaptable, comfortable and usable. These are known as Intelligent Vehicles.
This will be the sixth edition of the workshop, aiming at continuing the success of previous editions. The vast range of open challenges to achieve Safety and Security in Intelligent Vehicles (with or without connection with the Internet) is a fundamental reason that justifies the numerous research initiatives and wide discussion on these matters, which we are currently observing everywhere. Therefore, the workshop will keep its focus on exploring the challenges and interdependencies between security, real-time, safety and certification, which emerge when introducing networked, autonomous and cooperative functionalities.
- AI and Reliability Trends in Safety Critical Autonomous Systems on Ground and AirJyotika Athavale (Intel), Michael Paulitsch (Intel), Andrea Baldovin (Intel), Ralf Graefe (Intel), and Rafael Rosales (Intel)
- Reward Tuning for self-adaptive Policy in MDP based Distributed Decision-Making to ensure a Secure Mission PlanningMohand Hamadouche (Lab-STICC, CNRS), Catherine Dezan (Lab-STICC, CNRS), and Kalinka Regina Lucas Jauqie Castelo Branco (Universidade de Sao Paulo)
- The Quantitative Risk Norm - A Proposed Tailoring of HARA for ADSFredrik Warg (RISE Research Institutes of Sweden), Rolf Johansson (Autonomous Intelligent Driving), Martin Sanfridson (Volvo Technology AB), Mattias Brännström (Zenuity AB), Magnus Gyllenhammar (Zenuity AB), Martin Skoglund (RISE Research Institutes of Sweden) and Anders Thorsén (RISE Research Institutes of Sweden)
- Analysis of Cybersecurity Mechanisms with respect to Dependability and Security AttributesBehrooz Sangchoolie (Dependable Transport Systems, RISE Research Institutes of Sweden), Peter Folkesson (Dependable Transport Systems, RISE Research Institutes of Sweden), Pierre Kleberger (Dependable Transport Systems, RISE Research Institutes of Sweden) and Jonny Vinter (Dependable Transport Systems, RISE Research Institutes of Sweden)
- Exploring Fault Parameter Space using Reinforcement Learning-based Fault InjectionMehrdad Moradi (University of Antwerp and Flanders Make vzw), Bentley James Oakes (University of Antwerp and Flanders Make vzw), Mustafa Saraoglu (Technische Universitat Dresden), Andrey Morozov (Technische Universitat Dresden), Klaus Janschek (Technische Universitat Dresden) and Joachim Denil (University of Antwerp and Flanders Make vzw)
- Flexible Deployment and Enforcement of Flight and Privacy Restrictions for Drone ApplicationsNasos Grigoropoulos (University of Thessaly) and Spyros Lalis (University of Thessaly)
- Conceptual Design of Human-Drone Communication in Collaborative EnvironmentsHans Dermot Doran (Institute of Embedded Systems, ZHAW), Monika Reif (Institute of Applied Mathematics and Physics, ZHAW), Marco Oehler (Zurich University of Applied Sciences), Curdin Stöhr (Zurich University of Applied Sciences), and Pierluigi Capone (Centre for Aviation, ZHAW).
- A hierarchical fault tolerant architecture for an autonomous robotFavier Anthony (LAAS-CNRS, INPT ENSEEIHT - University of Toulouse), Messioux Antonin (LAAS-CNRS, INPT ENSEEIHT, University of Toulouse), Jérémie Guiochet (LAAS-CNRS,UPS, INPT, University of Toulouse), Jean-Charles Fabre (LAAS-CNRS, UPS, INPT, University of Toulouse) and Charles Lesire (ONERA/DTIS, University of Toulouse).
"Future Challenges in Safety and Security of Intelligent Vehicle"
- Mario Trapp (Fraunhofer IKS, Germany)
- Sibin Mohan (University of Illinois, USA)
- Miriam Gruber (BMW, Germany)
- Behrooz Sangchoolie (RISE, Sweden)
Workshop videos with Q&A
Session 1: AI and adaptive systems, and Session 2: Dependability and security analysis
Session 3: Architecture and deployment
Panel + Closing
15:00 – 18:00 (CEST).
Cross-Layer Soft-Error Resilience Analysis of Computing Systems
In a world with computation at the epicenter of every activity, computing systems must be highly resilient to errors even if miniaturization makes the underlying hardware unreliable. Techniques able to guarantee high reliability are associated to high costs. Early resilience analysis has the potential to support informed design decisions to maximize system-level reliability while minimizing the associated costs. This tutorial focuses on early cross-layer (hardware and software) resilience analysis considering the full computing continuum (from IoT/CPS to HPC applications) with emphasis on soft errors. The tutorial will guide attendees from the definition of the problem down to the proper modeling and design exploration strategies considering the full system stack (i.e., from circuit to software).
- Provide a deep understanding of the cross-layer impact of hardware faults on the full system
stack, taking into account all derating factors from technology (silicon) to software.
- Describe and analyze methodologies and tools for the evaluation of the resilience of each
system layer (i.e., circuit, microarchitecture, and software).
- Illustrate how specific approaches for resilience analysis working at different layers of the
system stack can be integrated to provide full system level analysis.
- Showcase the accuracy, strengths and weaknesses of the presented techniques.
Students, researchers and practitioners working on computing systems hardware and software design, with concerns about the impact of hardware faults on the full system level operation.
It is expected a basic understanding of computing systems hardware and software such as: logic design, computer architecture and microarchitecture, operating systems and programming. Some basic background on hardware defect mechanisms, fault and error modeling.
The tutorial is organized in an incremental manner. It starts with an introduction to reliability and cross-layer techniques followed by the main techniques applied at each abstraction level (e.g., circuits, architecture and software). The last part is focused on the most advanced concepts of stochastic cross-layer modelling, analysis and optimization. The agenda will be:
- Introduction – Basic Concepts, Terminology (30 minutes)
- Technology level resilience assessments (30 minutes)
- Microarchitecture level resilience assessments (30 minutes)
- Software level resilience assessments (30 minutes)
- Stochastic based approach for System level resilience assessments (30 minutes)
- Alberto Bosio, École Centrale de Lyon, France
- Stefano Di Carlo, Politecnico di Torino, Italy
- Alessandro Savino, Politecnico di Torino, Italy
- Dimitris Gizopoulos, University of Athens, Greece
- Ramón Canal, Universitat Politècnica de Catalunya and Barcelona Supercomputing Center, Spain
Tutorial video with Q&A
Into the Unknown: Unsupervised ML Algorithms for Anomaly-Based Intrusion Detection
One of the open challenges of past and recent systems is to identify errors before they escalate into failures. To such extent, most of the Error Detectors or enterprise Intrusion Detection Systems adopt signature-based detection algorithms, which consist of looking for predefined patterns (or "signatures") in the monitored data in order to detect an error or an ongoing attack. Data is usually seen as a flow of data points, which represent observations of the values of the indicators at a given time. Signature-based approaches usually score high detection capabilities and low false positive rates when experimenting known errors or attacks, but they cannot effectively adapt their behaviour when systems evolve or when their configuration is modified. As an additional consequence, signature-based approaches are not meant to detect zero day attacks, which are novel attacks that cannot be matched to any known signature. Moreover, when a zero-day attack that exploit newly added or undiscovered system vulnerabilities is identified, its signature needs to be derived and added as a new rule to the IDS.
To deal with unknowns, research moved to techniques suited to detect unseen, novel attacks. Anomaly detectors are based on the assumption that an attack generates observable deviations from an expected – normal – behaviour. Briefly, they aim at finding patterns in data that do not conform to the expected behaviour of a system: such patterns are known as anomalies. Once an expected behaviour is defined, anomaly detectors target deviations from such expectations, protecting against known attacks, zero-day attacks and emerging threats. To such extent, most of the anomaly detection algorithms are unsupervised, suiting the detection, among others, of unknown errors or zero-day attacks, without requiring labels in training data
The primary learning objectives of the tutorial are to demonstrate the capability of unsupervised learning algorithm to detect cyber-attacks and in particular zero-day attacks, and to instruct the attendees on the process to perform a well-crafted evaluation campaign.
In fact, after showing the current threat landscape as expanded by technical reports of agencies as ENISA, we will introduce anomaly detection, which is acknowledged as the most reliable answer to the detection of unknown errors or attacks. The participants will understand and use unsupervised algorithms that are particularly suited for anomaly detection, the main families and the differences in the way they decide if a data point is anomalous or normal. Participants will be involved in an hands-on session by using the RELOAD tool, which allows executing unsupervised anomaly detection algorithms and observing metric scores they provide on different datasets. This hands-on session, which can be conducted individually or in groups, will originate the final session which will constitute the final takeover of the tutorial, based both on participants activities and organizers’ experience in the domain.
The RELOAD tutorial targets anyone who is interested in the application of unsupervised ML algorithms for intrusion detection, with PhD students or young researchers as primary target audience. Consequently, we expect a remarkable amount of conference attendees to be interested in the topics of this tutorial, which targets beginners, with some content for intermediate. In fact, the tool to be used in the hand-on session will allow PhD students, researchers and practitioners who are starting to explore the discipline to get their first quantitative estimation of attack detection capabilities of algorithms, hiding implementation details which may be difficult to control at a first stage.
The tutorial will be composed by the following blocks.
- B1. Digression on the Current Threat Landscape (10% of tutorial time). Starting from public reports e.g., ENISA, we will describe the current state of cyber-attacks.
- B2. Anomaly-Based Intrusion Detection (15% of tutorial time). This part highlights some key terms and components that will be used in the rest of the tutorial, alongside with its role in detecting intrusions.
- B3. Unsupervised Algorithms and their Characteristics (10% of tutorial time). We will introduce some of the most common algorithms to be used for unsupervised anomaly detection.
- B4. Presentation of the RELOAD Tool (15% of tutorial time): This part will let the audience understand what the RELOAD tool offers, and how to use the RELOAD tool for executing unsupervised algorithms.
- B5. Hands-On Session (40% of tutorial time): the attendees can use the tool to perform intrusion detection on public attack datasets that are previously downloaded by the organizers and shared with the slides.
- B6. Wrap-up and Final Discussion (10% of tutorial time): Results obtained during hand-on session will be discussed together with the audience, originating final discussions. We will prepare spare material for enriching the discussion, expanding on already existing studies.
- Tommaso Zoppi, University of Florence, Italy
- Andrea Ceccarelli, University of Florence, Italy
- Andrea Bondavalli, University of Florence, Italy
Tutorial videos with Q&A
Into the Unknown: Unsupervised ML Algorithms for Anomaly-Based Intrusion Detection - Part 1
Into the Unknown: Unsupervised ML Algorithms for Anomaly-Based Intrusion Detection - Part 2
The InterPlanetary File System and the filecoin network
The InterPlanetary File System (IPFS) is a peer-to-peer content-addressable distributed file
system that seeks to connect all computing devices with the same system of files. It is an
and a global community of millions of users.
IPFS resembles past and present efforts to build and deploy Information-Centric Networking
approaches to content storage, resolution, distribution and delivery. IPFS and libp2p , which is
the modular network stack of IPFS, are based on name-resolution based routing. The resolution
system is based on Kademlia DHT and content is addressed by flat hash-based names. IPFS
sees significant real-world usage, with over 250,000 daily active network nodes, millions of end
users and wide adoption by several other projects in the Decentralised Web space, but not only.
An adjacent project to IPFS, which was also masterminded and is also being developed within
Protocol Labs (the umbrella company of IPFS and libp2p) is filecoin . Filecoin is a cryptocurrency
that supports a decentralised storage and delivery network. Storage and retrieval miners are
rewarded according to their contribution to the network and the mechanics of filecoin secure the
network against malicious activity
The main objective of this tutorial is to let researchers, developers, and users understand IPFS
and the capabilities it provides.
More specifically, participants will:
- Understand how IPFS brings content addressing as a core primitive for data distribution
- Learn how to use CIDs (content identifiers) to find content and interpret what the content
- Learn how to create custom data structures using IPFS and its underlying data format,
IPLD (InterPlanetary Linked Data)
- Understand how libp2p bring process addressing as a core primitive for P2P and
The attendees do not need to have prior knowledge of IPFS, libp2p or filecoin and basic
knowledge and understanding of core networking and network security principles will be
adequate in order to follow along.
- Understanding how IPFS deals with files (60 mins)
- Solving distributed networking problems with libp2p (60 mins)
- The lifecycle of data in IPFS and filecoin (40 mins)
- Developing Apps with the IPFS API (20 mins)
- David Dias, Peer-2-Peer Software Engineer at Protocol Labs, (Palo Alto, CA and Lisbon,
- Dr. Ioannis Psaras, EPSRC Fellow and University Lecturer (Assistant Professor) at
University College London and a Research Scientist at Protocol Labs.
Tutorial videos with Q&A
The InterPlanetary File System and the filecoin network - Part 1
The InterPlanetary File System and the filecoin network - Part 2
The InterPlanetary File System and the filecoin network - Part 3
Top-level research is a highly competitive environment: funding; recruiting; publishing; impact … If you move in the first division, academia is like a premier league, and top researchers are high-level competition athletes. Is that too stressing? Where is the fun? Depends on the perspective. There is no unique recipe, but I’ll share my own experience and hope to show that it can be a unique life, if you do the right things.
If you manage the balance between freedom, self-responsibility, and perseverance, chances are you will go far, and have moments you'll never forget. How far? Well, if you are aiming for the gold, the nice secret of this talk is that excellence … is about details.
Paulo Esteves-Veríssimo is a professor and FNR PEARL Chair at the University of Luxembourg FSTM and SnT, and Head of the CritiX lab (https://wwwen.uni.lu/snt/research/critix). He is adjunct Professor of the ECE Dept., Carnegie Mellon University. Previously, he has been a professor of the Univ. of Lisbon (PT). He is the representative of UNILU-SnT in ECSO, the European Cyber Security Organisation, and member of its Scientific & Technical Committee (STC). He was Chair of the IFIP WG 10.4 on Dependable Computing and Fault-Tolerance and vice-Chair of the Steering Committee of the IEEE/IFIP DSN conference.
He is Fellow of the IEEE and Fellow of the ACM, and associate editor of IEEE Trans. on Emerging Topics in Computing (TETC). He is currently interested in architectures, middleware and algorithms for resilient modular and distributed computing, in areas like: SDN-based infrastructures; autonomous vehicles from earth to space; digital health and genomics; or blockchain and cryptocurrencies. He is author or co-author of over 200 peer-refereed int’l publications and
co-author of 5 books. Check his pubs on GSC.
- Safeguarding Data Consistency at the EdgeClaudio Correia (Universidade de Lisboa, Portugal)
- Depending on HTTP/2 for Privacy? Good Luck!Gargi Mitra (IIT Madras, India)
- Towards Practical Privacy-Preserving Collaborative Machine Learning at a ScaleRania Talbi (INSA-Lyon, France)
- What Exactly Determines the Type?Inferring Types with ContextLigeng Che (Nanjing University, China)
- Impact of geo-distribution and mining pools on blockchains: a study of EthereumPaulo Mendes da Silva, INESC-ID & IST. U. Lisboa, Portugal
- CanvasMirror: Secure Integration of Third-Party Library in WebVR EnvironmentJiyeon Lee, KAIST, Korea
- A Framework for Risk Assessment in Augmented Reality-equipped Socio-technical SystemsSoheila Sheikh Bahaei, Malardalen University, Sweden
Videos of the sessions with Q&A
Welcome and Keynote