Category: Talk Ideas

23rd of June, at 14h00, Karthik Pattabiraman will give a presentation entitled“Building Error Resilient Machine Learning Systems from Unreliable Components” 
Location: G4.1

Bio
Karthik Pattabiraman is a Professor of Electrical and Computer Engineering at the University of British Columbia (UBC). He received his MS and PhD in computer science from the University of Illinois at Urbana Champaign (UIUC) in 2004 and 2009, and spent a postdoctoral year at Microsoft Research (MSR), Redmond before joining UBC in 2010. His research interests are in dependability, security, and software engineering. Karthik has won multiple awards such as the Inaugural Rising Star in Dependability Award, 2020, from the IEEE and the IFIP, the distinguished alumnus award from the University of Illinois (UIUC), CS department, 2018, and multiple UBC-wide awards for excellence in research and mentoring. Together with his students and collaborators, he has published over 100 papers, many of which have received distinguished paper awards at venues such as DSN and ICSE. He is a distinguished contributor of the IEEE computer society, a distinguished member of the ACM, and the vice-chair of the IFIP Working Group on dependable computing and fault-tolerance (WG 10.4).  A more detailed biography may found at: https://blogs.ubc.ca/karthik/about/full-bio/

Abstract
Machine Learning (ML) has increasingly been adopted in safety-critical systems such as Autonomous vehicles (AVs) and industrial robotics. In these domains, reliability and safety are important considerations, and hence it is critical to ensure the resilience of ML systems to faults and errors. Hardware faults such as soft errors are becoming more frequent in commodity computer systems due to the effects of technology scaling and reduced supply voltages. These faults can lead to ML systems malfunctioning, and cause safety violations. Further, errors in the  training data have been widely observed even in mature  training datasets, and these can lead to significant degradation of accuracy in ML algorithms. Therefore, there is a compelling need to protect ML systems from both hardware faults and training data errors. 
In this talk, I’ll present  some of the work we’re doing in my group to ensure the dependability of ML systems in the presence of hardware faults and training data errors. For the former, we introduce Ranger, an automated transformation for Deep Neural Network (DNN)-based systems that can filter out the hardware faults that are likely to have the most impact in the DNN. For the latter, I’ll present the use of ensemble-based techniques, and show that they outperform most other techniques proposed in the ML community for dealing with training data errors.  This is joint work with my students and colleagues at UBC, as well as with industry collaborators.  

14th of June at 16h00, Sandino B. Jardim and Gustavo Callou  will give two short presentations, to promote discussion on two relevant ongoing or disruptive topics. Afterwards, there will be a social gathering where everyone can talk freely on whatever subjects they like.
Location: G4.1

Sandino B. Jardim – “Evolving Benchmarking Methodologies for SDN Controllers” 

Bio
Sandino B. Jardim is a professor at the Federal University of Mato Grosso (UFMT). He holds an M.Sc and a Ph.D. in Computer Science from the Federal University of Goiás (UFG) and the Federal University of Rio Grande do Norte (UFRN), respectively. Currently, he is conducting postdoctoral research at CISUC in the field of benchmarking methodologies for SDN Controllers.

Abstract
The emergence of Software-Defined Networking (SDN) brought about a paradigm shift in the networking landscape. SDN controllers quickly emerged as potential systems under test (SUT) for benchmarking, with most works focusing on traditional system quality attributes such as performance and scalability. As SDN evolved as a key enabler for subsequent innovations like network function virtualization, IaaS, IoT, and 5G, its role expanded and gained significant importance for the success of these technologies. Benchmarking methodologies also transformed, incorporating additional quality attributes such as dependability and reliability. Given the shifting landscape of the SDN domain and the growing need for comprehensive benchmarking, there is a demand for research on updating benchmarking methodologies for SDN controllers. This research aims to address the evolving requirements and advancements in SDN technology, ensuring that benchmarking practices keep pace with emerging challenges and opportunities.

Gustavo Callou – “An Approach to Integrate Performance Assessment, Dependability, Energy Efficiency, and Security Applied to Cloud Computing” 

Bio
Gustavo Callou is an associate professor at the Federal Rural University of Pernambuco (UFRPE), Brazil. He has a Ph.D. in Computer Science from the Federal University of Pernambuco (UFPE), Brazil, with a split-site doctoral program at University of Wuppertal, Germany, in Dependability and Performance Evaluation. Callou is the leader of the Performance and Data Analysis Lab (PEDAL) at UFRPE, and his key research interests include Reliability Analysis, Fault-Tolerant Computing, Performance Engineering, Sustainability, and Cloud Computing.

Abstract
Energy efficiency represents an important factor that can be used to reduce the environmental impact. Forecasting required resources in conjunction with an energy optimization strategy represents a challenging problem because of cloud computing environments’ dynamic nature and varying workload characteristics. In the literature, dynamic migration of virtual machines (VMs) between servers is commonly proposed to increase security, although it has essential disadvantages, such as migration cost and performance degradation. This project aims to propose an approach to assist in managing resources in virtualized private clouds, and that seeks to optimize the number of servers to meet the requirements of dynamic workloads. This research will also seek to quantify the impact of security policies (e.g., VM migration) on performance, power consumption, and dependability. To evaluate this proposed strategy, experiments will be conducted in the laboratory with a private cloud environment and models will be proposed to assist in optimization and identify the best approach to deal with these conflicting requirements.

24th of May, at 16h00, Chriss IT. Leong will give a presentation entitled“Translating Natural Language Requirements to Formal Specifications: A Study on GPT and Symbolic NLP” 
Location: G4.1

Bio
Chriss IT. Leong is a Chief Information Security Officer, EHR, Department of Information Security, Health Bureau. Is is also a Phd Student at CISUC, Department of Informatics Engineering, University of Coimbra. He has a Master and Bachelor in Computer Science from National Yang Ming Chiao Tung University, Taiwan. His research interests focus on Software Engineering, Formal Specification and Software Verification.

Abstract
Software verification is essential to ensure dependability and that a system or component fulfils its specified requirements. Natural language is the most common way of specifying requirements, although many verification techniques such as theorem proving depend upon requirements being written in formal specification languages. Automatically translating requirements into a formal specification language is a relevant and challenging research question, because developers often lack the necessary expertise. We consider the application of natural language processing (NLP) to address such research question. Two distinct approaches are proposed to formalise natural language requirements: a symbolic method and a GPT-based method. Both methods are evaluated with respect to their ability to generate accurate Java Modeling Language (JML) from textual requirements, and the results show good promise for automatic formalisation of requirements.

17th of May, at 16h00, Zoran Budimac will give a presentation on the research conducted at the Department of Mathematics and Informatics of the University of Novi Sad
Location: G4.1

Bio
Zoran Budimac was born in 1960. in Serbia. The academic rank title of full professor got from University of Novi Sad in 2004. Currently works at Faculy of Sciences, Univesity of Novi Sad, Serbia.
During his career he has published 280 research papers, 16 books and presented his work on 180 international conferences.Currently, he is the head of one laboratory and the head of chair of computer scince. he is also  a member of University counsil and member of management committee of an international journal.
Currently, his scientific interest are in various aspects of software quality.

10th of May, at 16h30, Zane Bicevska will give a presentation on the research conducted at the Faculty of Computing of the University of Latvia
Location: G4.1

Bio
Dr. Zane Bicevska is a professor at the University of Latvia, and a dean of the Faculty of Computing. She holds a PhD in Computer Science from the University of Latvia and a master degree in Economics from the J.W.Goethe University of Frankfurt am Main (Germany). She has also studied computer science at the University of Buffalo, USA. Dr. Bicevska has more than 20 years of experience in the field of computer science and software engineering, and she is also acting as an enterpreneur in IT industry.Dr. Bicevska is an expert in software engineering, software quality, software testing, and project management. She teaches project management and software testing to bachelor and master level students at the University of Latvia, and  has published more than 40 research articles in international journals and has presented her work at various international conferences.

26th of April, at 16h00, Charles Gonçalves will give a presentation entitled“Intrusion Injection for Virtualized Systems: Concepts and Approach” 
Location: G4.1

Bio
Charles F. Gonçalves received a B.Sc. and an M.Sc. degree in computer science from the Federal University of Minas Gerais (UFMG) Brazil. He is currently pursuing a Ph.D. degree in Informatics Engineering at the University of Coimbra (UC). He is a Software and System Engineering (SSE) Group member in the CISUC and also from the SPEC Research Security Benchmarking Working Group. His interests include Virtualization Security, Security Benchmarking, Data Processing, and Software Development. 

Abstract
Virtualization is drawing attention due to countless benefits, leaving  Hypervisors with the paramount responsibility for performance, dependability, and security. However, while there are consolidated approaches to assessing the performance and dependability of virtualized systems,  solutions to assess security are very limited. Key difficulties are evaluating the system in the presence of unknown attacks and vulnerabilities and comparing the security attributes of different systems and configurations when an intrusion occurs. 
In this talk, we will discuss the concept  of intrusion injection for virtualized environments, which consists of directly driving the system into the erroneous states that mimic the ones resulting from actual intrusions (in the same way errors are injected to mimic the effects of residual faults).

20th of April, at 13h00, Antônio Fröhlich will give a presentation entitled“SmartData: a Data-centric Approach to the Design of Safety-Critical Systems” 
Location: G4.1

Bio
Antônio Augusto Fröhlich is a full professor at the Federal University of Santa Catarina (UFSC), where he leads the Software and Hardware Integration Laboratory (LISHA) since 2001. He holds a PhD in Computer Engineering from the Technical University of Berlin (2001) and was a visiting researcher at the University of Paderborn (2007), at the University of California, Irvine (2016) and at the University of Luxembourg (2017). He has coordinated several research and innovation projects in connected, secure and intelligent embedded systems, including the ALTATV Digital TV Platform, the CIA² research network on Smart Cities and the Internet of Things, the Smart Campus project at UFSC and Smart Grid projects with partner industries. In 2022, he was a founding member of UFSC’s Research Center for Cyber-physical Systems Security (SecCPS), currently acting as Executive Director.

Abstract
Data is at the core of the design of modern Safety-Critical Systems. Data is no longer only sensed and processed in the context of the control loops of such systems. It is also secured, stored, and transmitted for the sake of the decision-making processes required for higher levels of autonomy. The task-centered strategies traditionally used to design critical systems consistently support scheduling analysis and verification of tasks execution times, as long as periods, deadlines and execution time estimates are known, but mostly ignore the flow of data across the various components in the system and often assume that data generation time is constant and can be fully encapsulate in the execution time of tasks. A Data-driven approach to the design of such systems can more promptly accommodate requirements such as data freshness, redundant data sources, operational safety, and AI-readiness. It also facilitates the design of mechanisms to monitor, and eventually override, non-deterministic components such as the neural networks commonly used in autonomous systems. In this talk, we present the SmartData concept, which was conceived at LISHA/UFSC to handle these design issues. The talk addresses aspects of domain decomposition, scheduling, communication, formal property specification and verification, and component overriding. The Autonomous Vehicle being built at UFSC lays a case background for the talk.

14th of April, at 14h00, Alexandru Iosup will give a presentation entitled“Massivizing Graph Processing: The Science, Design, and Engineering of a Complex Ecosystem” 
Location: A5.4

Bio
Dr.ir. Alexandru Iosup is a full professor at Vrije Universiteit Amsterdam (VU), a high-quality research university in the Netherlands. He is the tenured chair of the Massivizing Computer Systems research group at the VU and visiting researcher at TU Delft. He is also elected chair of the SPEC-RG Cloud Group. His work in distributed systems and ecosystems includes over 150 peer-reviewed articles with high scientific impact, and has applications in cloud computing, big data, scientific and business-critical computing, and online gaming. His research has received prestigious recognition, including membership in the (Young) Royal Academy of Arts and Sciences of the Netherlands, the Netherlands ICT Researcher of the Year award, and a PhD from TU Delft. His leadership and innovation in education led to various awards, including the prestigious Netherlands Higher-Education Teacher of the Year. He has received a knighthood for cultural and scientific merits. Contact Alexandru at A.Iosup@vu.nl or @AIosup, or visit http://atlarge.science/aiosup

Abstract
Wherever we turn, our society is digital. Digital data and digitalized processes are becoming critical for science and engineering, decision-making and business-critical operations, and online education and gaming. An emerging building block of digitalization in the 21st century, graph representation of both real and digital states, is becoming common in practice. At societal and even global scale, these digital elements depend, often transparently from the perspective of their clients, on the effective inter-operation of efficient computer systems into large ecosystems, managed largely without a developer and even client input. However successful until now, we cannot take these ecosystems for granted: the core does not rely on sound principles of science and design, and there are warning signs about the scalability, dependability, and sustainability of engineering operations. This is the challenge of massivizing computer systems, and, as the focus here, of massivizing graph processing. 
In this talk, inspired by the recently started EU project Graph Massivizer [1], and by our experience with distributed computer systems for over 15 years, we focus on understanding, deploying, scaling, and evolving graph-processing ecosystems successfully. We explain GraphMassivizer’s ambitious, comprehensive research program that aims to develop a serverless, sustainable, graph processing platform, with applications to sustainable ICT infrastructure, AI/ML, FinTech, and Industry 4.0. We posit that we can address the fundamental challenges of massivizing graph processing by focusing on computer ecosystems rather than merely on (individual, small-scale) computer systems. We showcase diverse results in massivizing graph processing, including (1) a reference architecture, and a resource management and scheduling framework, that can span the computing continuum, (2) the design and development of graph processing engines, and (3) the development of a benchmark and performance analysis framework that led to a fundamental performance result. 
The session will be interactive, so bring over your questions, comments, and curiosity. We will have a team with diverse professional backgrounds ready to answer them. Thank you for joining!
[1] Graph Massivizer website: https://graph-massivizer.eu/

12th of March at 16h00, Anamta Khan and Jessica Castro  will give two short presentations, to promote discussion on two relevant ongoing or disruptive topics. Afterwards, there will be a social gathering where everyone can talk freely on whatever subjects they like.
Location: G4.1

Anamta Khan – “A Machine Learning driven Fault Tolerance Mechanism for UAVs’ Flight Controller” 

Bio
Anamta Khan is a PhD student in the SSE group at DEI. She obtained her Master’s and Bachelor’s degrees in Computer Science from the Institute of Business Administration (IBA) in Karachi. Her research focus since 2021 has been on creating a safer airspace with maximum and efficient utilization of drones. She utilizes fault injection in a simulation-based environment to identify vulnerabilities and their impact, and then uses machine learning-based techniques to address them.

Abstract
Unmanned Aerial Vehicles (UAVs), similar to other robotic systems, are susceptible to various hazards (e.g., software or hardware failures or security attacks) that may hinder mission completion and compromise safety by violating the separation minima (i.e., the minimum distance that must be maintained between UAVs in order to ensure safe and efficient operations). To address this issue, this paper proposes a new machine learning-based fault-tolerant mechanism for UAV flight controllers that tolerates GPS-related hazards and improves safety. Machine learning models were built using 884,410 data records from 1,985 flight logs collected from the PX4 public repository. The trained models are used to predict the expected position of the UAV during a mission, and separation minima are used as a threshold to detect the GPS hazards by comparing it with the distance between two consecutive position values. When a hazard is detected (i.e., the distance is higher than separation minima), the predicted values by machine learning models are fed into the position estimator (i.e., EKF) of the flight controller. To evaluate the effectiveness of this approach, validation experiments were conducted on several realistically defined missions while being exposed to different types of failure conditions (e.g., Noise and GPS failure), both with and without using the proposed fault-tolerant mechanism. The results show a remarkable reduction in the safety violations from 94 to 1 (violation of the separation minima counts), indicating a significant improvement in safety. Additionally, the proposed mechanism demonstrated a notable improvement in the distance covered and duration of the flight mission in failure conditions, demonstrating its ability to mitigate faults effectively. These findings support the effectiveness of the proposed fault tolerance mechanism in enhancing UAV mission performance by improving safety and tolerating issues caused by GPS.

Jessica Castro – “Attack Detection in Microservice Applications” 

Bio
Jessica Castro is a PhD candidate in Informatics Engineering at the University of Coimbra, Portugal. She received a BSc. in Computer Science and an MSc in Informatics from Universidade Federal da Paraíba, Brazil. Her research interests include attack detection, microservices/containerised applications, and self-adaptive systems.

Abstract
A microservices-based architecture decreases the complexity of developing new systems, making them highly scalable and manageable. However, its distributed nature, the high granularity of services, and the large attack surface increase the challenge of securing those systems. We will present the challenges of monitoring and identifying attacks in microservice applications at runtime and how we aim to use logic scoring of preference to calculate scores that allow identifying possible attacks.