GPU-Programming with CUDA

Short course on GPU programming. The instructor will be Marius Brehler from Dortmund University (Germany), and a course summary is available here

Schedule:

– Session 1: Monday, May 20,   14:30 – 17:30,          Meeting room of Alan Turing building;

– Session 2: Tuesday, May 21,   14:30 – 17:30,          Meeting room of Alan Turing building;

– Session 3: Wednesday, May   22, 14:30 – 17:30,    Room to be defined.

Two seminars on Robotics and Automotive

Title: Effective and User-friendly Specification of Multi-Robot Missions
When: Wednesday 24, April 15:00-16:00
Where: Aula seminari, blocco 0 

Abstract: Mobile robots are increasingly used in everyday life to autonomously realize missions such as exploring rooms, delivering goods, or following certain paths for surveillance. The current robotic market is asking for a radical shift in the development of robotic applications where mission specification is performed by robotic users that are not highly qualified and specialized in robotics or ICT. To this aim, we proposed a Domain Specific Language (DSL) that enables non-technical users to specify missions for a team of autonomous robots in a user- friendly and effective way. The talk will present how we developed such DSL, it semantics, and how we evaluated it.

Bio: Sergio García is a PhD student at the Computer Science and Engineering department of Gothenburg University. His research interests comprise model-driven software development and software architecture for robotics. His research is currently involved with the Co4Robots European Project, from where he often collaborates with industry.

Title: On Interfaces to Support Agile Architecting in Automotive
When: Monday 29, April 15:00-16:00
Where: Aula seminari, blocco 0 

Abstract: In large-scale agile automotive companies, practitioners struggle with creating and evolving an architecture when developing complex and safety-critical systems. A key issue is the trade-off between upfront planning and flexibility to embrace change. In particular, the coordination of interfaces is an important challenge, as interfaces determine and regulate the exchange of information between components, subsystems, and systems, which are often developed by multiple teams. In a fast-changing environment, interfaces between teams can provide the sufficient stability to align software or systems, while maintaining a sufficient degree of autonomy. In this talk, we present an exploratory case study with an automotive OEM to identify characteristics of different interfaces, from non-critical interfaces that can be changed frequently and quickly, to those that are critical and require more stability and a rigorous change process. We identify what dimensions impact how interfaces are changed, what categories of interfaces exist along these dimensions, and how categories of interfaces change over time. We conclude with suggestions for practices to manage the different categories of interfaces in large-scale agile development.

Bio: Rebekka Wohlrab is an industrial PhD student at the Computer Science and Engineering Department of Chalmers University of Technology and Systemite AB. In her research, she focuses on large-scale automotive companies aiming for more agile ways of managing documentation. Her work is funded by the Wallenberg AI, Autonomous Systems and Software Program.

GSSI Seminar by Adrian Rutle

When: Wednesday, April 17, 2019, 15:00 

Where: Library Room, GSSI

Title: Model Repair with Reinforcement Learning
Seminar by Prof. Adrian Rutle, Department of Computing, Mathematics and Physics – Western Norway University of Applied Sciences

Abstract:Model Driven Engineering is an emerging branch of Software Engineering used to handle complex and evolving software systems. The industrial demand is quite high and with the increasing use of Model-Based Development in many domains (e.g., Automotive, Web Applications, Business processes), models are becoming core artifacts of modern software engineering processes. With this, it comes the necessity to keep models free of errors and in case errors happen, to repair them reasonably. This seminar will present our proposed approach to model repair with focus on these three topics:

Model repair methodologyWhen models increase in size and complexity, they tend to become hard to keep free from mistakes.In this seminar, we propose reinforcement learning algorithms as a step forward to achieve repair of broken models allowing both customization of results and automation at the same time. As a proof of concept, we have built a bridge between model repairing and reinforcement learning, designed a system of rewards to support customization of results and implemented a concrete scenario of model repairing. Although we have applied our approach to models defined in the Eclipse Modeling Framework (EMF), the approach is general enough to be extended to other modeling frameworks.We have validated our research by repairing a large set of broken models generated with a mutation tool.

Personalization, customization and transfer learning In the model repairing field there are tools providing automatic repairing of models.Some of them support user personalization, but none stores information from the repairing process.Reusing experience from past repairings would help to avoid duplicated calculations when facing repeated personalization preferences.Our tool PARMOREL, uses reinforcement learning algorithms as a step forward to achieve repair of broken models allowing both personalization of results and automation at the same time. We propose transfer learning as an approach to reuse the experience learned from each execution of PARMOREL.We have built a theoretical approach to support transfer learning in PARMOREL.Also, we have validated our research by repairing a model using different sets of preferences and studying how our tool’s performance evolved when reusing the experience from each repairing.


Using metrics for calibration of rewardsAs part of the seminar, moreover, we present a proposal for integrating quality assurance into PARMOREL.We describe an architecture that would allow PARMOREL to learn to automatically repair models with the highest quality possible.

Seminar by Luca Berardinelli on “Model-Driven Engineering in Practice at Braintribe: tools, challenges, and collaborations”

When: Tuesday, April 16th, 11AM

Where: Meeting Room, “Alan Turing” building

Title: Model-Driven Engineering in Practice at Braintribe: tools, challenges, and collaborations.   


Summary: Braintribe (https://www.braintribe.com) is a SME founded in 2005, with around 70 employees, located in Vienna, Zurich, Belgrade, Bratislava, London, Sao Paolo, and Frankfurt. Braintribe provides two main products, Tribefire and Datapedia that are horizontal, cross-domain modeling/metamodeling and asset (e.g., COTS) sharing platforms based on models and data normalisation approaches.

Their core component, namely Cortex, defines a new technical space in the MDE domain. This seminar aims at showing the platform for the first time to the academy and to MDE experts to discuss tool capabilities and limitations, shape challenges for a wise adoption of MDE principles and best practises, and promote future collaborations.

Speaker:Luca Berardinelli is a PhD in Computer Science. He works as a Technical Lead at BrainTribe, with particular focus on Model/Data integration, with special focus on Smart Cities and Smart Factories.

He received his Ms and PhD degrees in Computer Science from the University of L’Aquila, Italy. He has been a postdoc at the Univ. of L’Aquila (2011-15) and TU Wien (2015-18) working on Model-Driven Engineering (non-functional analyses, context and uncertainty modeling) in many European (PLASTIC,   VISION, PRESTO,  CRAFTERS, U-TEST) and national projects (MAE4ASE/SysML4Industry), participating as representative of his research departments to intermediate and final review meetings.

LinkedIn: https://www.linkedin.com/in/lucaberardinelli/ Research Gate:  https://www.researchgate.net/profile/Luca_Berardinelli

Seminars Prof. Carlo Fischione, Monday 15 and Tuesday 16, from 2 pm to 6 pm

Prof. Carlo Fischione (Royal Institute of Technology – KTH, Sweden) will give two classes on Monday 15 and Tuesday 16, from 2 pm to 6 pm in Aula Seminari DISIM (Alan Turing Building), on the following topics:

Title: Fundamentals of Machine Learning over Networks

Lecturer: Prof. Carlo Fischione

Abstract: This course covers fundamentals of machine learning over networks (MLoNs). It starts from a conventional single-agent setting where one server runs a convex/nonconvex optimization problem to learn an unknown function. We introduce several approaches to address this seemingly, simple yet fundamental, problem. We introduce an abstract form of MLoNs, present centralized and distributed solution approaches to address this problem, and exemplify via training a deep neural network over a network. The course covers various important aspects of MLoNs, including optimality, computational complexity, communication complexity, security, large-scale learning, online learning, MLoN with partial information, and several application areas. As most of these topics are under heavy researches nowadays, the course is not based on a single textbook but builds on a series of key publications in the field. 

Bibliography

[1]        Bubeck, Sébastien. “Convex optimization: Algorithms and complexity.” Foundations and Trends in Machine Learning, vol. 8, no.3-4 (2015): 231-357.
[2]        L. Bottou, F. Curtis, J. Norcedal, “Optimization Methods for Large-Scale Machine Learning”, SIAM Rev., 60(2), 223–311.
[3]        Boyd, Stephen, et al. “Distributed optimization and statistical learning via the alternating direction method of multipliers.” Foundations and Trends in Machine learning 3.1 (2011): 1-122.
[4]        Goodfellow, Y. Bengio, A. Courville, “Deep Learning”, MIT press 2016
[5]        Jordan, Michael I., Jason D. Lee, and Yun Yang. “Communication-efficient distributed statistical inference,” Journal of the American Statistical Association, 2018.
[6]        Smith, Virginia, et al. “CoCoA: A general framework for communication-efficient distributed optimization.” Journal of Machine Learning Research 18 (2018): 230.
[7]        Alistarh, Dan, et al. “QSGD: Communication-efficient SGD via gradient quantization and encoding.” Advances in Neural Information Processing Systems. 2017.
[8]        Schmidt, Mark, Nicolas Le Roux, and Francis Bach. “Minimizing finite sums with the stochastic average gradient.” Mathematical Programming 162.1-2 (2017): 83-112.
[9]        Boyd, Stephen, et al. “Randomized gossip algorithms,” IEEE Transactions on Information Theory, 2006.
[10]     Scaman, Kevin, et al. “Optimal algorithms for smooth and strongly convex distributed optimization in networks,” ICML, 2017.

Lectio Magistralis at the GSSI on Software Heritage by Stefano Zacchiroli

Schedule: 24 April, 11.30 – 12.30 
Place: GSSI – Main Lecture Hall, v. Francesco Crispi 7

Speaker: Stefano Zacchiroli
IRIF, Université Paris Diderot / Inria / IRILL, Paris, France
https://upsilon.cc/~zack/

Title: “Software Heritage: source code analysis at the scale of the world”

Abstract: The Software Heritage project has assembled the largest existing archive of publicly available software source code and associated development history, for more than 5 billion unique source code files and 1 billion unique commits, coming from more than 80 million development projects.

In this talk we will review the project background, current status, and future directions with a focus on its research applications. In particular, we will discuss (1) using the Software Heritage archive as a research object, allowing to study software evolution, patterns, and development dynamics at an unprecedented scale; and (2) how Software Heritage is a fundamental building block for open science, and how it contributes to increasing accessibility and reproducibility of scientific source code. We will conclude the talk with a brief overview of the research challenges that need to be overcome to fulfill the Software Heritage mission.

Collaborative and Confidential Information Sharing and Analysis for Cyber Protection

Speaker: Rogerio de Lemos, University of Kent, UK
When: Wednesday, April 03rd
Where: Sala Riunioni di Matematica, Coppito I building at: 2:30 PM

Abstract:
Analysing cyber threat information (CTI) provides organisations with valuable intelligence about which of their systems are being attacked, and who is attacking them. If organisations could pool their CTI then it is quite likely that other, possibly low level, distributed attacks would be identified. But organisations are not usually willing to share  their CTI because of the confidential and private information that it  contains. If organisations could be re-assured that the sharing would have minimal consequences, according to their risk and trust profiles, then this should be beneficial to the them and the entire community. The EC H2020 C3ISP project is enabling organisations to share their CTI by allowing them to specify Data Sharing Agreements (DSAs), and by enforcing these DSAs either on the organisation’s own premises or in a trusted third party service provider, prior to the analysis. This talk will described the various technologies that comprise the C3ISP infrastructure, and how they can be distributed and integrated in order to allow organisations to share their CTI in a trustworthy manner. The four validating pilot projects, involving CERTs, ISPs, multinationals and SMEs will also be briefly described.  

Speaker short bio:
Rogerio de Lemos is a senior lecturer in the School of Computing at the University of Kent since 1999. In 2009, he was an invited assistant professor at the University of Coimbra in Portugal. Previously to joining Kent, he was a Senior Research Associate at the Centre for Software Reliability (CSR) at the University of Newcastle upon Tyne. His 
research interests are on software engineering for self-adaptive systems, architecting dependable and secure systems, insider threats, and resilient AI.

Towards Compositional Transformations for Dependability Analysis of Evolving and Reconfigurable Systems

Speaker: Kristof Marussy, Fault Tolerant Systems Research Group, University of Budapest
When: Tuesday, March 19, 3:15PM
Where: Aula 1.7 (Coppito 1)
Title: Towards Compositional Transformations for Dependability Analysis of Evolving and Reconfigurable Systems

Abstract:
Evaluation of the reliability and performance of variable and adaptable architectures as well as the co-evolution of architectures and analyses remains difficult. Evolving system structure often requires repeated analysis with significant computational costs and manual intervention. Therefore analysis methods need to prepare for evolving systems and reconfigurations. In my research, I focused on the construction of analysis models by change-driven, forward only view transformations to efficiently propagate architecture changes to stochastic Petri net analysis models. I present a domain-specific language to describe the view transformations in terms of Petri net fragments and their interconnections in a fully compositional way, which means individual transformations can be combined without re-engineering. I also present a transformation engine suited for such view transformations: in addition to source– and target-incremental execution, fully compositional transformations can be run even in the face of validation errors in the source or target models. Lastly, I present the description and analysis of adaptation strategies by semantic integration of high-level models and analyses in phased-mission stochastic models, which is enabled by the incremental execution of the view transformation.

Discovering complex individual and social behavior patterns via Social network Analysis

Speaker: Prof. Pasquale De Meo, Università di Messina
When: Wednesday, February 6, 2019
Where: AULA A1.4 COPPITO ZERO
Title: Discovering complex individual and social behavior patterns via Social network Analysis

Thanks to the availability of user profiles and records of activity, online social network analysis is relevant to discover complex individual and social behavior patterns.
The emergence of trust between users of online services is one of the most important phenomena, but it’s also hard to detect in records of users’ interactions, and even harder to replicate by abstract, generative models.
Seminario 1: February 6, 2019 at 11:00
In this seminar we report our results on the emergence of “trusted” users (over time) by studying the evolution of topological and centrality measures in the network mapping trust relations among members of an Online Social Network.

Seminario 2: February 6, 2019 at 14:30
This seminar illustrates how social network analysis tools can be used to discover reputable users in online communities: given a “who-trusts-whom” network in which users can create trust relationships as well as review items and evaluate posted reviews, we investigate whether the most central actors in the network are also highly reputable, i.e., if they are likely to provide helpful reviews to the benefit of all community members.

Seminar “Facing Uncertainty in Complex Cyber-Physical System Design”

Speaker: Malina Software Corp. (Canada), Simula Research Laboratory (Norway), Monash University (Australia)
When: Thursday, February 7, 2019
Where: room Alan Turing, Coppito Zero (Blocco 0)
Title: Facing Uncertainty in Complex CPS Design

ABSTRACT: The unprecedented complexity of many modern-day cyber-physical systems (CPS) requires changes in how we design and develop such systems. Traditional methods were typically based on the assumption that a capable and responsible design team will identify all potential uncertainties in a proposed design and, through careful and systematic analysis, reduce or even eliminate the consequent risk prior to committing to a given design alternative. However, experience has amply demonstrated that, once a system exceeds a certain threshold of complexity, it is unrealistic to expect that even the best and most experienced design team can anticipate and accurately uncover all possible sources of uncertainty and accurately assess their consequences. For instance, due to their sheer number and complexity, it is very difficult to predict potential interference between independently defined system functions (this is sometimes referred to as the feature interaction problem). Consequently, given that we cannot hope to fully eliminate uncertainty in such systems, we must learn how to incorporate and deal with it in the design process.
To that end, it is first necessary to develop a proper understanding of uncertainty: what it is, how it is manifested, and how it can be represented. In this talk, we describe one conceptual model of uncertainty, the UTaxonomy, which was developed as part of the European H2020 “UTest” project. Although this project is focused on the problem of testing CPS in the presence of uncertainty, the conceptual model was designed to be general and is likely to be useful in other uncertainty-related research. To illustrate how such a model can be applied in practice, we briefly explain how it is being used to identify and describe uncertainties when specifying requirements.

 

Speaker Bio: Bran Selić is President of Malina Software Corp., a Canadian company that provides consulting services to corporate clients and government institutions worldwide. He is also Director of Advanced Technology at Zeligsoft Limited in Canada, and a Visiting Scientist at Simula Research Laboratories in Norway. In 2007, Bran retired from IBM Canada, where he was an IBM Distinguished Engineer responsible for setting the strategic direction for software development tools. Currently, he is also an adjunct professor at Monash University and the University of Sydney in Australia. With over 40 years of practical experience in designing and implementing large-scale industrial software systems, Bran has pioneered the application of model-based engineering methods and has led the definition of several international standards in that domain, including the widely used Unified Modeling Language (UML). In 2016, he was presented with a lifetime Career Award by the steering committee of the IEEE/ACM MoDELS conference in recognition of his contributions to model-driven technologies and practice.