ERASMUS MUNDUS - mscdesem.ie PRinter2.pdfErasmus Mundus Double MSc in Dependable Software Systems (DESEM). This is the first return of the DESEM summer school to Maynooth, since the

MSC IN DEPENDABLE SOFTWARE SYSTEMS

ERASMUS MUNDUS

DESEM

SUMMER SCHOOL

• July 4th to July 8th, 2016

• Venue - Eolas Building, Maynooth University

• In association with University of St Andrew’s, Scotland and Universite de Lorraine, France

JULY 2016

ERASMUS MUNDUS PAGE 2

SPONSORS & INDUSTRIAL CONTRIBUTORS


WELCOME

Welcome to the fourth summer school of the Erasmus Mundus Double MSc in Dependable Software Systems (DESEM). This is the first return of the DESEM summer school to Maynooth, since the initial DESEM summer school was held here in July 2013.

We look forward to an exciting program of academic activities, coupled with a number of social and cultural events. The academic activities will focus on presentations by DESEM scholar students from each of the participating universities: The University of St Andrews (Scotland), The Université de Lorraine (France) and of course Maynooth University (Ireland). I encourage everyone to avail of this wonderful opportunity to find out about and interact with your fellow students, particularly those from the other DESEM institutions.

As well as student presentations, we have a number of guest academics giving talks on their area of expertise and shedding new light on issues related to the dependability of software systems. Industry participation will also take place and will include some presentations from industry experts.

I hope this will be an enjoyable and productive week, showing us the breadth of accomplishments and achievements in the area of dependable software systems.

DR DIARMUID O’DONOGHUE DESEM PROGRAMME DIRECTOR


SCHEDULE

MONDAY 09:30 - 10:00 Welcome Meeting MORNING STUDENT PRESENTATIONS 10:00 - 10:25 Phattara Wangrungarun Semi-autonomous object tracking navigation for drones with cameras and simple elementary motions Supervisor: Michael Weir 10:25 - 10:50 Milena Dyle From UML behavioral diagrams to PROMELA and back Supervisor: Dr Juliana Bowles 10:50 - 11:25 COFFEE BREAK 11:25 - 11:50 Worakarn Isaratham Procedural Generation of Tactical RPG Battle Systems Supervisor: Ian Miguel 11:50 - 12:15 Junyi Zhang Automated Library System with Image Recogntion for Local Primary Schools Supervisor: Thomas Naughton 12:15 - 12:40 Ahmed Abdelghany A Software Tool for Assessing the Humanitarian Preparedness of Regions in OpenStreetMap Supervisor: Dr Peter Mooney 12:40 - 13:40 LUNCH AFTERNOON 13:00 - 14:30 Juliana Bowles, St Andrew’s University Making a difference: applications of formal methods in healthcare

14:30 - 15:30 Inthraporn Aranyanak, Burapha University Computation and Neuroscience: synergies and inter-dependencies 15:30 - 16:00 COFFEE BREAK 16:00 - 16:30 Didier Galmiche, Universite de Lorraine An Epistemic Separation Logic 16:30 - 17:00 Brian Smyth, SQS


SCHEDULE

TUESDAY MORNING STUDENT PRESENTATIONS 09:00 - 9:25 Suhyun Cha Project title: From Clinical Pathways to Event Structures for Formal Verification Supervisor: Dr Juliana Bowles 09:25 - 9:50 Sebastian Duque Project title: Verification of programs written in Elixir Supervisor: Dr Juliana Bowles 9:50 - 10:15 Devrim Gunyel Project title: Identifying Signs of Diabetic Retinopathy Using Deep Learning Supervisor: Prof Barak Pearlmutter 10:15 - 10:45 COFFEE BREAK 10:45 - 11:10 Rohit Sharma Project title: Learning to Forgetting: Incorporating Biology Inspired Forgetting Mechanisms into Machine Learning Algorithms

Supervisor: Dr. Ognjen Arandjelović 11:10 - 11:35 Leonardo Pais Cardoso Project title: An intelligent Air Quality Data Alert Monitoring Tool Supervisor: Dr Peter Mooney 11:35 - 12:00 Hager Ali Project title: Near Analogy and Text Similarity with Dr Inventor Supervisors: Dr Y Abgaz, Dr D O’Donoghue & Dr D Hurley 12:00 - 12:45 LUNCH AFTERNOON 12:45 - 17:00 Field Trip to SAP


SCHEDULE

WEDNESDAY MORNING STUDENT PRESENTATIONS 09:30 - 09:55 Mahmoud Abulebdeh Project title: Software system to gather, fuse and integrate geo-referenced data from different Social Media Newtorks Supervisor: Dr Peter Mooney 09:55 - 10:20 Anders Candasamy Project title: Model-based development of a Hemodialysis Machine Supervisor: Prof Dominique Méry 10:20 - 10:45 Andrea Balogh Project title: Computational analysis of games that require computational thinking Supervisor: Dr Tom Naughton 10:45 - 11:15 COFFEE BREAK 11:15 - 11:40 Keylor Chaves Project title: NavWell: Experimental software for studying spatial navigation abilities in healthy people and patients. Supervisor: Joe Duffin 11:40 - 12:05 Fady Zaki Project title: Specifying Software Architectures in Java Supervisor: Dr Dharini Balasubramaniam 12:05 - 12:30 Weichung Shaw Project title: Formal Verification of Raft Supervisor: Stephan Merz 12:30 - 14:00 LUNCH


SCHEDULE

AFTERNOON GUEST SPEAKERS 14:00 - 15:00 Prof Ait Ameur Ontologies as formal models to make explicit domain knowledge. Application to model annotation. 15:00 - 15:15 COFFEE BREAK 15:15 - 16:45 Dr Stephen Brown, Maynooth University Java Instrumentation


THURSDAY MORNING STUDENT PRESENTATIONS 09:00 - 9:25 Juan Karsten Project title: Accurate and Automatic Localization of Human vertebrae from X-ray Computed Tomography Scans. Supervisor: Dr Ognjen Arandjelović

09:25 - 09:50 Dmitry Smorodinnikow Project title:Evaluation of “Subject-Verb-Object” Inferences in Blended Spaces Supervisors:Dr Y Abgaz, Dr D O’Donoghue & Dr D Hurley 09:50 - 10:15 Akulina Lyapunova Project title: A Plugin for refactoring Event-B Models Supervisor:Dr James Power 10:15- 10:40 Guillermo Reyes Guzman Project title: A system for camera tracking in indoor environments Supervisor: Dr John McDonald 10:40- 11:00 COFFEE BREAK 11:00 - 11:25 Rodrigo Rodrigues de Carvalho Project title: Visualizing and Analyzing Software Artefacts and their Relationships. Supervisor: Dr Dharini Balasubramaniam 11:25 - 11:55 Gokberk Kocak Project title: Decade Analysis of an Audio Database using Deep Learning Techniques Supervisor: Dr Joe Timoney 12:00 - 12:45 LUNCH AFTERNOON Cultural Trip to Glendalough and social evening at

Leopardstown racecourse


FRIDAY MORNING 09:30 - 11:00 Summer School Questionnaire Completion

Exit Interviews for 2nd Year Students.

Poster Set up

11:00 - 11:30 COFFEE BREAK INDUSTRY SPEAKERS 11:30 - 12:00 Dave Cleary, Huawei Research 12:00 - 12:30 SAP PAST PUPIL 12:30 - 13:00 Juan Jose Mendoza On the verification of UML models using NLP AFTERNOON 13:00 - 14:00 Poster Session and Lunch 14:00 - 14:30 Prize Giving/Speeches 15:00 - 16:00 Russell Library Tour, NUIM


JULIANA BOWLES

GUEST SPEAKER

BIOGRAPHY:

Juliana Bowles (née Küster-Filipe) is a senior lecturer at the School of Computer Science, University of St Andrews, Scotland. She has research interests in formal methods and dependability, logics and models for true-concurrency, and foundations of modelling languages. She is currently applying her foundational work to the healthcare domain, and has an EPSRC-funded project on "Automated Conflict Resolution in Clinical Pathways", as well as a collaboration with Brazil funded by the Royal Academy of Engineering under its Newton Research Collaboration Programme on the use of quantitative analysis techniques to improve the efficiency of complex healthcare processes. Bowles received her PhD (Dr.rer.nat) from the Technical University of Braunschweig, Germany, in 2000.

TUTORIAL: MAKING A DIFFERENCE: APPLICATIONS OF FORMAL METHODS IN HEALTHCARE

This talk presents different logic-based approaches with potential applications in healthcare. Examples include ways to detect and resolve conflicts in the treatment of patients with multiple chronic conditions; and to predict and compare the effect of cancer treatments with a particular emphasis on elderly patients. Ongoing and

future work will be highlighted.


INTHRAPORN ARANYANAK

TEACHING SCHOLAR

BIOGRAPHY: Dr.Inthraporn Aranyanak (Joop) studied Computer Science at Maynooth University, completing her PhD in 2015. Her PhD thesis involved the study of braille reading. She developed a high-resolution finger tracking system to track readers’ finger movements. She obtained a master’s degree in linguistics from Chulalongkorn University, Thailand in 2006 and worked as a linguist for two years in King Mongkut University of Technology Thonburi. In 2008, she worked in the Department of Computer Science at Maynooth University as a researcher, before starting her PhD there in 2009. She is currently a lecturer in the College of Research Methodology and Cognitive Science, Burapha University, Thailand. Her research interests are in the areas of language and cognition with a specific focus on reading.

TUTORIAL: COMPUTATION AND NEUROSCIENCE: SYNERGIES AND INTER-DEPENDENCIES This talk will explore the relationships between software engineering and neuroscience. While at first there may not seem to be many points of contact between these two disciplines, I hope to highlight a number of informative overlaps that potentially can inform the development of large-scale, realistic simulations of brain function of the type currently funded under the EU's one billion euro Human Brain Project. In addition, there are useful lessons to be learned from the resilience of brain function to failure and damage for the field of robust and reliable software development. I will illustrate these connections with examples from the literature and highlight some further potentially useful connections that might be made between the fields of neuroscience and software engineering. Note that I will provide a general introduction to the relevant neuroscience concepts at the beginning of the talk.

https://www.facebook.com/pages/King-Mongkuts-University-of-Technology-Thonburi/105707952796519

https://www.facebook.com/pages/King-Mongkuts-University-of-Technology-Thonburi/105707952796519


DIDIER GALMICHE

GUEST SPEAKER

BIOGRAPHY: Didier Galmiche is Professor in Computer Science in Université de Lorraine (UL) in France. He is the scientific leader of the research group TYPES (Logic, Proof theory and Programming) in LORIA UMR 7503, Nancy and his main research areas are: substructural and non classical logics; type theory, typed lambda calculi for programming and proving; resource logics for modelling and proving systems properties; semantics and proof theory; proof-search methods, decision procedures. He is the director of the Master programme in Computer Science in UL and also the local coordinator of the Erasmus Mundus DESEM programme.

TALK: AN EPISTEMIC SEPARATION LOGIC We present an Epistemic Separation Logic, called ESL, that allows us to consider epistemic possible worlds as resources that can be shared or separated, in the spirit of separation logics. After studying the semantics and the expressiveness of this logic, we provide a tableau calculus with labels and resource constraints that is sound and complete and then also study counter-model extraction. It is a joint work with Jean-René Courtault and Hans van Ditmarsch (Université de Lorraine, LORIA UMR 7503, Nancy).


JUAN JOSE MENDOZA

GUEST SPEAKER

BIOGRAPHY: Juan Jose Mendoza Santana, better known as Juanjo, is a Mexican Software Engineer and entrepreneur. He is currently pursuing a PhD degree in Computer Science at the University of St Andrews in the UK and also acts as the honorary CTO of Krieger Mexico. He completed the Erasmus Mundus Master in Dependable Software Systems in 2014. His latest research is related to the use of NLP and Formal Methods in Software Engineering, specifically in the verification of UML models. He also has several years of wide experience in the development of systems used in industry, from banking software to mobile application for civic hacking. Juanjo loves playing volleyball, travelling, learning new stuff, and hanging around his friends.

TALK: ON THE VERIFICATION OF UML MODELS USING NLP UML models are the predominant software representation in Model Driven Development. Hence, the correctness of such models with respect the requirements, and the consistency within their parts are crucial to build dependable systems. Existing tools check for internal consistency of different diagrams, or take advantage of OCL to partially verify behavioural constraints. However, none of these tools provide means to prove that the generated UML models satisfy the natural language requirements. This issue is addressed in this research by combining formal methods with NLP. Our approach includes the development of appropriate semantics that enable the verification of the models against the requirements represented as a knowledge-base.


PROF YAMINE AIT AMEUR

GUEST SPEAKER

BIOGRAPHY: Yamine Ait Ameur is Full Professor since 2000 at ENSEEIHT (Ecole Nationale Superieure d'Electronique, Electrotechnique, Informatique, Hydraulique et Telecommunications) in Toulouse (France). Previously he was at ENSMA (National School of Mechanics and Aeronautics) in Poitiers (France) between 2002 and 2011, and he has been the head of LISI/ENSMA (Laboratory of Industrial and scientific computer science at ENSMA). He got his HDR (Habilitation to conduct research) in 2000, and his PhD in Computer Science in 1992 at ENSAE-SUPAERO. His Research topics concern: i) Formal methods for validation and verification, ii) Ontololgy based modeling and ontology based databases, iii) Application domains: embedded systems, interactive systems, semantic web, PDM databases, etc. Two main important directions characterize his research activities. On the one hand the fundamental aspects through the use of formal modeling techniques based on refinement and proof, explicit formalization of semantics using formal ontology models. On the other hand, practical aspects, through the development of operational applications, to validate the proposed approaches. Embedded systems in avionics, engineering, interactive systems, CO2 capture are some of the application domains targeted by this work.

TALK:

ONTOLOGIES AS FORMAL MODELS TO MAKE EXPLICIT DOMAIN KNOWLEDGE. APPLICATION TO MODEL ANNOTATION. Critical systems are running in heterogeneous domains. This heterogeneity is rarely considered explicitly when describing and validating processes. Handling explicitly such domain knowledge increases design models robustness due to the expression and validation of new properties mined from the domain models. This talk discusses an approach to enrich design models describing complex information systems with domain knowledge. We use ontologies to model such domain knowledge. Design models are annotated by references to domain ontologies. The resulting annotated models are checked. It becomes possible to verify domain-related properties and obtain strengthened models. The approach is deployed for two design model development approaches: a Model Driven Engineering (MDE) approach and a correct by construction formal modelling one based on refinement and proof using Event-B method. This talk presents first the notion of ontologies and their formalisation within set theory. Next, the Event-B method is overviewed. The design models we consider are represented by state transitions systems. Then, we show, on case studies how ontologies can be used to annotate design models and thus how these models are strengthened after annotation.


STEPHEN BROWN

GUEST SPEAKER

BIOGRAPHY: Stephen Brown is a Senior Lecturer in Computer Science at NUIM. Following his graduation in Electronic Engineering (BA,BAI) and Computer Science (MSc) from TCD he spent 10 years in industry, working on and managing embedded networking R&D for Digital Equipment Corporation. He returned to academia as a Research Fellow in TCD on the ADVANCE ESPRIT project. Subsequently he moved to NUIM where he introduced the M.Sc. in Software Engineering, was heavily involved as Head of Department in the introduction of the undergraduate degree course in Computer Science and Software Engineering (CSSE), and was Director of the Innovation Value Institute (IVI) for a number of years. In addition to a wide variety of undergraduate topics, his current lecturing speciality at M.Sc. level is Software Testing. He is a researcher in the Callan Institute, where his main research topic is Wireless Sensor Networks and F1 in the SFI/Connect Center.

TALK:

JAVA INSTRUMENTATION Java code can be instrumented for a number of purposes: but mainly to measure test coverage and performance. In this workshop we will look at a number of different ways to instrument Java code: from high level tools, to low level bytecode manipulation. If you want to follow along during the workshop, you might want to make sure you have the following installed on your laptop: Java 7 (JDK), Javassist, jvisualvm, JaCoCo, Java ASM 2.0. Also, Eclipse will be used for running some of the examples. Source code of the examples will be made available during the workshop.


DAVID CLEARY - INDUSTRY SPEAKER

BIOGRAPHY: David Cleary is currently the CTO for the Service Provider Operation SPO business unit in Huawei based in Stockholm Sweden. SPO division has the product and service end-2-end responsibility for the management of cloud, NFV and SDN portfolio. David also co-chairs the TM Forum Frameworx Architecture and Steering Team (eTOM, TAM, SID and TIP). Prior to joining Huawei in February 2011, David held many senior technical architect roles in Ericsson. David was the founding member of Ericsson’s Network Management Research Centre in 2001, charged with the responsibility to direct the technology evolution of Ericsson OSS portfolio. This role included responsibility for overall OSS/BSS standardisation activity across the Ericsson group. David is the author of numerous publications and patents in these fields.

TALK:

TRUTH OR DARE “DEPENDABLE OPEN SOURCE SOFTWARE IN THE ERA OF NFV AND SDN” After decades of standardisation and millions of hours in agreements in paper standards, the telecommunication and IT industry is facing its biggest challenge – the emergence of open source implementation- the movement towards software defined infrastructure running on Commercial off-the-shelf (COTS) hardware. Telecommunication industry has traditionally followed rigorous standards for stability, protocol adherence and quality. A side effect of this is the slow speed of development. The second failing is the reliance on specific hardware, e.g. application-specific integrated circuits (ASICs). The Over the top communication provider (such as Google Talk, Skype, Netflix) has spurred service providers to look for ways to disrupt the status quo. In this talk I will focus on the key challenges that our industry is facing. I will use examples from the current open source project being proposed(openstack, opNFV, openO, OSM et al) as the basis for the next generation of Network functions virtualization (NFV) and Software-defined networking (SDN).

https://en.wikipedia.org/wiki/Application-specific_integrated_circuit

https://en.wikipedia.org/wiki/Application-specific_integrated_circuit

https://en.wikipedia.org/wiki/Google_Talk

https://en.wikipedia.org/wiki/Google_Talk

https://en.wikipedia.org/wiki/Skype

https://en.wikipedia.org/wiki/Netflix

https://www.sdxcentral.com/nfv/

https://www.sdxcentral.com/nfv/

https://www.sdxcentral.com/sdn/resources/what-the-definition-of-software-defined-networking-sdn/


BRIAN SMYTH - INDUSTRY SPEAKER

BIOGRAPHY: Current Role: Head of Functional Test Service Company: SQS Ireland Ltd. Brian Smyth is a senior leader with over 10 years experience working for a Global Quality Assurance Consultancy across the Banking, Financial Services, Insurance, Telecoms and Utilities sectors. During this time, Brian has successfully delivered multiple large scale multi-million euro projects, managing geographically dispersed teams in both Agile and Waterfall environments. Brian's area of expertise is in Quality Management and Quality Process Improvement having completed numerous Quality Strategy Audits for key organisations. In addition, Brian has planned, defined and rolled out QA Centres of Excellence for these organisations to deliver an increased level of efficiency, traceability, risk management and auditability to the Programme/Project Lifecycle while capturing key performance metrics to ensure continuous improvements. In his current role, Brian is responsible for leading a team of over 100 consultants from both a technical and career management perspective, in addition to being accountable for all aspects of Testing Service Innovation, Service Delivery and Service-line revenue.


PHATTARA WANGRUNGARUN - STUDENT

BIOGRAPHY: A web/front-end developer from Thailand, generally likes Ruby and JavaScript, loves gaming, manga (comic), both reading and drawing, hanging out and drinking, casual sports and winter sports.

PROJECT:

SELF-AUTONOMOUS OBJECT TRACKING NAVIGATION FOR DRONES WITH CAMERAS AND SIMPLE ELEMENTARY MOTIONS A Parrot’s AR.Drone 2.0 quad-copter has a built-in front camera, which is capable of being used as a sensor for tracking a target object and provides a numerial information about the target which the drone sees at each certain time, such information can be its current width, height, or colour.

The project uses this idea to create a semi-autonomous system which navigates the drone according to the information from the target. The target is to be presented clearly visible in front of the drone for a certain distance before the system starts navigating the drone, hence the term semi-autonomous.

However, available motion controls for the drone from the library that the system builds on top are simple and limited. For example, one can only specify a numerial speed for turnings, not an angle, or specify a numerial speed for moving forward, not a distance. The project aims to create more sophisticated navigations based on these elementary motions. Such navigations are simple approach-to and retreat-from the target, and orbiting around the target.



MILENA DYLE - STUDENT

BIOGRAPHY: I was born in Tirana, Albania. Currently I am following my final year of DESEM master in University of St Andrews, Scotland. I graduated from University of Tirana in Computer Science and spent my first year of DESEM master in Université de Lorraine, France. During my studies I have always been proactive in participating in different activities, competitions, youth exchanges and sports. In my 20s I started a par-time job as software developer at IKUBINFO, a software engineering company in Albania, followed by a fulltime position after graduations. During this period I have been involved on software development of large projects for the Albanian government. At the same time, I was offered a position as a visiting teacher assistant for the seminars of Algorithmic at Faculty of Sciences, University of Tirana. My main focus is software engineering, software modeling, software testing and formal methods application to deliver reliable and high quality software. Moreover I like learning new cultures, languages and always in love with nature.

PROJECT: FROM UML BEHAVIORAL DIAGRAMS TO PROMELA AND BACK UML is a widely standard language which plays an important role in describing and specifying the behavior of a system and its environment. One of the main features of a dependable software system is to prove the system correctness which is usually done using formal methods. SPIN is a model checker which proves the correctness of a model specified in a language called PROMELA. SPIN searches the state space of the designed system and returns information whether a desired property is fulfilled or not, associated with a trace when it fails in the later case. What is interesting is that formal method is good, but it requires system engineers’ prior knowledge in math and logic. This dissertation aims to provide a software tool that allows software engineers design dependable systems using UML and check system properties with no prior knowledge in formal methods by using SPIN as a model checker, and PROMELA as a model specification language behind the scenes. The tool provides an interface to include UML representations diagrams and an interface to write properties the system should or should not meet.



WORAKARN ISARATHAM - STUDENT

BIOGRAPHY: Growing up in the suburbs of Bangkok, Worakarn Isaratham discovered his interest in programming during middle school. Enrolling to the department of Computer Engineering at Chulalongkorn University was an obvious choice at the time. After completing his bachelor's degree, he worked as quality assurance engineer at Reuters Software (Thailand), gaining specialisations in automated testing and nonfunctional testing. After 4 years, he converted from the testing side to the development side, still at the same company, working on several projects using Java and .NET technologies. A year later he left the company for a small local startup company Qlovr, to work as an iOS developer. Then the opportunity to study in DESEM program came up and he could not let it slip away. In his free time he likes to sit down and read a good book. His reading list includes fictions, nonfictions, and increasingly more and more graphic novels. Appreciating good music from all over the world is his other passion.

PROJECT: PROCEDURAL GENERATION OF TACTICAL RPG BATTLE SYSTEMS Procedural generation is a method that enables game content to be generated algorithmically and automatically, rather than manually. Common application of this technique is currently being used for low-level content such as textures, vegetation, clouds, game maps, road networks, or puzzles. This work attempted to apply procedural generation to tactical role-playing games (TRPG). Central to games in this genre are turn-based battles between game characters, which resemble complex chess-like board games. The complexity of the battles arises from the variety of game characters, and the combat actions they can perform. Crucial to the enjoyment of the game is the balance between all the available choices of characters and actions, which we collectively call the battle system. Calibrating the system to be in balance is usually done manually. Instead, we created a prototype TRPG battle, and a procedure that algorithmically generates a well-balanced battle system for the prototype. An objective function was defined, based on self-play game AI. We then used genetic algorithm to optimise the objective function, to find the battle system that fits the objective best. Finally, we had human volunteers play the game running on various battle systems, to validate that the procedurally-optimised one is in good balance.



JUNYI ZHANG - STUDENT

BIOGRAPHY: My name is Junyi Zhang, I was born and grew up in China. When I was an undergraduate, I participated in various societies and helped organize many events such as monthly Cosplay and on-campus cleaning campion. I got my Bachelor Degree of Engineering in Computer Science in Harbin Institute of Technology in 2014. During my bachelor studies, I became an exchange student in Korea. This exchange experience helped me open up my mind and I really enjoyed the international environment. So after graduation, I joined this EM DESEM program. Last year I finished my first year study in St Andrews and my thesis topic was User defined rule-based Filing System with Periodic Execution. Last summer I joined Toshiba Medical Visualization System Europe Ltd in Edinburgh as a software engineer intern. Now I am studying in Maynooth and my dissertation is about Automated Library Category System with Image Recognition.

PROJECT: AUTOMATED LIBRARY SYSTEM WITH IMAGE RECOGNTION FOR LOCAL PRIMARY SCHOOLS Nowadays, barcode-based digital systems has made it easy to manage books in modern libraries. However, these systems are not suitable for small-scale libraries in primary schools, where budget is tight and children’s books without ISBN are normal. As required by a local primary school, we proposed and implemented an automated library system based on image recognition, which enables pupils to borrow or return books by simply taking a photo of the book cover, without assistance from the librarian. This is a three-year open-source project. As the initiators of the project, we kicked off this project by interviewing parent representative and principal of the school. Based on the analysis of their requirements, we designed, implemented and tested the automated library system. We also evaluated the performance and usability of our system through experiments and an online survey. The evaluation results showed that the system is a satisfying prototype but further optimization is required. Therefore, we explicitly addressed future work for upcoming developers, i.e. improvement of image recognition, upgrade of GUI, and involvement of pupils in the system evaluation if possible.



AHMED ABDELGHANY - STUDENT

BIOGRAPHY: Ahmad Abdelghany is an Egyptian software developer. He received his bachelor‘s degree in computer systems from Ain Shams University in Cairo. He worked as a software engineer for various multinational telecom companies including Asset Technology Group and Orange. He also holds multiple programming certificates including Certified Scrum Master (SCM), Certified Web Component Developer (SCWCD), Certified Java Programmer (SCPJ), and Certified SQL Expert (OSE).

Ahmad plans to join Dublin’s vibrant tech scene after finishing his master’s degree in Dependable Software Systems. Apart from coding, he also enjoys reading, traveling and all kinds of sports.

PROJECT: A SOFTWARE TOOL FOR ASSESSING THE HUMANITARIAN PREPAREDNESS OF REGIONS IN OPENSTREETMAP OpenStreetMap (OSM) is a collaborative mapping project that maintains a spatial database of geographical information for the entire world. During a humanitarian disaster OSM can be used by emergency services and can be of great help to them on the ground. OSM is often referred to as Wikipedia of maps. This project aims at developing a robust tool to analyze OSM data for a given region and assess humanitarian preparedness of that region. This assessment could be used by Humanitarian agencies to have an idea about the completeness and usability of OSM data for the effected region. The tool could also give information to the OSM community on missing information and features that should be considered for mapping.



SUHYUN CHA - STUDENT

BIOGRAPHY: I studied Electronic Engineering in Chungnam National University of South Korea for my BSc and in Korea Advanced Institute of Science and Technology (KAIST) for my MSc. Later, I worked in the IT department for six years. And, finally, I decided to study computer science because it affects more on everyday life and industries than hardwares do as time goes by and I would like to be work on in this field. I have been accepted for DESEM program and got full scholarship. I have spent my first in NUIM. I am currently studying at University of St. Andrews for my second year.

PROJECT: FROM CLINICAL PATHWAYS TO EVENT STRUCTURES FOR FORMAL VERIFICATION There is an increasing number of people with two or more longterm medical conditions, with women and the elderly often more vulnerable. Even though there are standardized descriptions and medical procedures such as clinical guidelines for the treatment of individual chronic diseases (e.g.,in the UK, NICE has more than 350 pathways documented), it is a challenge to detect existing conflicts between these guidelines, because clinical processes themselves are very complex. In concurrent system modeling, event structures can be used to describe specific computation flows, and contain formal notions to capture causality, conflict and concurrency between events. In this project, we explore a mapping from clinical pathways to LESs (Labelled Event Structures). Moreover, we show an approach to construct a formal model (a LES) of a given clinical pathway, where the latter is described in BPMN (Business Process Modeling Notation), and how to link it to constraint solvers for conflict detection. We evaluate the approach using actual examples of clinical pathways.



SEBASTIAN DUQUE - STUDENT

BIOGRAPHY: My name is Sebastián and I’m from Medellín, Colombia. I got my BEng Systems Engineering from Universidad EAFIT in Medellín and worked for almost three years as a Software Engineer for a financial services company. After that I became a DESEM student in 2014 and started in Maynooth. At the previous summer school in Nancy I was disappointed that, although I had taken some French lessons, “je ne parle pas français” was my most used sentence. Thinking I could leave that feeling behind me I went to St Andrews for my second year, only to be confronted with the Scottish accent. Up until now I’ve travelled across four continents to about two dozen countries and lived in five of them – I’m eager to increase the count!

PROJECT: VERIFICATION OF PROGRAMS WRITTEN IN ELIXIR The evolution of computer hardware has long shifted from increasing the clock frequency in single-core processor machines to increasing the number of processors resulting in multi and many-core machines. The logical effect for software, in order to benefit from these changes, is to abandon the single execution thread pattern and to embrace concurrency/parallelism. Different concurrency models have been developed to tackle this. Our project focuses on the actor model introduced in 1985 by Gul Agha and now supported in programming languages like Erlang (1993), Scala (2004) and Elixir (2012) which we conducted our study on. We aim to examine the Elixir ecosystem to identify and provide mechanisms to reason about the correctness of distributed and concurrent programs. Looking at both informal (unit testing frameworks and property based testing) and formal (model checking) verification approaches and techniques like Design by Contract we have smoothed off some rough edges with existing tools and evaluated the extent to which they can be applied in a successful manner.



DEVRIM GUNYEL - STUDENT

BIOGRAPHY: She was born in 1986 in Turkey. Decided to be a scientist when she was 15 when someone offered her to be a theatre actress. She realised that whatever she would do, there should be mathematics in it. On the other hand, it didn’t stop her to act for 10 years and being a professional swimmer for 8 years till her 20s. After graduating as BSc. Computer Engineering degree, she started to work as junior consultant project manager for call centre technology companies. She worked for 4 years and realised that planning other engineers’ job and train them is not her thing and decided to be one of them. She started MSc. Computer Engineering in Turkey and met with Bioinformatics. Now, she is trying to figure out how to deal with academy after 5 years without programming. In addition, struggle to survive in academy, she works in human rights organisations and organising events about nationalism and politics, cooking, gardening and dream about being a scientist one day. PROJECT:

IDENTIFYING SIGNS OF DIABETIC RETINOPATHY USING DEEP LEARNING Diabetic Retinopathy is a visual impairment disease as a result of diabetes mellitus. Visual impairment caused by diabetic retinopathy(DR) can be cured if detected in earlier stage of DR in patients. In this work, we try to automatically detect signs of DR by using deep neural networks. For this purpose, we train deep neural network by using one of the widely used publicly available datasets and test with other public datasets. Experimental results show that deep learning techniques can be promising in the task of automatic detection of DR.



ROHIT SHARMA - STUDENT

BIOGRAPHY: Rohit Sharma is a second year DESEM student currently studying at University of St. Andrews, Scotland. He holds a Bachelor degree in Computer Science and Engineering from Panjab University, India. He is a former software engineer with 4 years of experience in areas of IT Operations, Research & Development, White box testing and automation. His areas of interest include machine learning, distributed systems and software architecture. Apart from coding, Rohit enjoys reading about nature, history and finance.

PROJECT: LEARNING TO FORGETTING: INCORPORATING BIOLOGY INSPIRED FORGETTING MECHANISMS INTO MACHINE LEARNING ALGORITHMS In the current era, where everything is connected, data is being generated at an ever increasing rate. Machine learning has transformed the way we analyze data to find hidden insights and enabled us to quickly and automatically produce models that can deliver faster and more accurate results. There is mounting evidence that forgetting plays an important role the human and animal learning process. Yet, not much work has been done in the field of machine learning to mimic forgetting irrelevant facts. The existing works in this area are overly simplistic and superficial.

This project is a step towards implementing a novel forgetting mechanism and incorporating it into simple baseline prediction algorithms. We also attempt to evaluate this novel approach by analyzing the improvement in prediction accuracy for baseline algorithms.



LEONARDO PAIS CARDOSO - STUDENT

BIOGRAPHY: Leonardo is a first year Erasmus master student at Maynooth University and is expected to continue his studies at the University of St Andrews. He is graduated in Computer Engineering from Universidade Federal do Rio de Janeiro (UFRJ), one of the most prestigious universities in Brazil, and finished his degree ranking among the 15% best students. During his undergraduate studies, he dedicated his time to research, being able to develop analytical, critical-thinking and problem solving skills. As a result he published 5 papers on International and Brazilian conferences in the fields of autonomous systems and virtualisation. Leonardo was also awarded a fully funded Scholarship by the Brazilian government to study one year at the University of Southampton in UK as a visiting student. In UK he also had the opportunity to work as a Software Engineer at ASV, a company dedicated to build autonomous surface vessels. At ASV he was able to develop his communication skills and understand more about the industry. Apart from academic life, Leonardo spends his free time helping friends, travelling and playing computer games.

PROJECT: AN INTELLIGENT AIR QUALITY DATA ALERT MONITORING TOOL The Irish Environmental Protection Agency (EPA) monitors the air quality in 30 stations all over Ireland in real-time. This data is used to inform the population about air quality because high concentrations of some pollutants can be harmful. Due to technical complexity of the monitoring network there can be issues with the data provided from the monitoring stations' sensors. If these are not detected early and correctly, false information about air quality can be given to the public. Currently, the detection of issues within the monitored data is performed mostly manually. This can be tedious and error prone, given the large number of stations and pollutants monitored.

In order to increase the performance, efficiency and accuracy of the detection, an intelligent air quality data monitoring software tool has been developed in this project. Detection is based on a number of measures including guidance from European directives for air quality. Whenever the tool detects an issue or problem an alert is sent to the EPA team informing them of the issue. This allows the team to identify a fix and resolution to the issue in a timely manner. While this tool is not intended for public use it serves an important purpose "behind the scenes" of ensuring that the public are always provided with the highest quality and most accurate air quality data.



HAGER ALI - STUDENT

BIOGRAPHY: I am a software developer who is currently studying MSc in Dependable Software Systems at Maynooth University. I graduated from the Faculty of Computer and Information Sciences, Ain Shams University, Cairo, Egypt with a first class honors BSc degree. During my senior year, I developed a collaborative graduation project “Multi-touch interactive screen” where I utilized the academic background of 4 years to add something new based on integrating human computer interaction with image processing techniques. The project awarded the first position in IEEE GOLD competition and got excellent marks with high recognition from professors and industrial figures. After graduation, I continued my education and got an opportunity to work for eSpace, which is a reputable software company in MENA region. My years of experience in the industry have led me to be exposed to various models and work on diverse projects. One of my favorite achievements during work is building the first Arabic MOOC platform for enhancing education in the region. My current philosophy is that reinventing the wheel is a complete waste of time, and we should utilize the right and convenient models to avoid ad hoc solutions and achieve more robust applications for making people’s lives better.

PROJECT: NEAR ANALOGY AND TEXT SIMILARITY WITH DR INVENTOR Finding semantic similarities between textual information has been one of the hot topics in natural language processing and cognitive computing. Although there are many diverse applications of text similarity in the software industry, their computational models are limited in terms of unlocking the metaphoric and analogical similarities. Our research aims to overcome this limitation by using an existing analogy-based model (Dr Inventor) that uncovers creative bisociations between documents. That is, for a given set of documents, we utilize Dr Inventor to detect near analogies between these documents to show the indirect similarity between these documents. Furthermore, we identify the small differences between similar texts. This thesis extends Dr Inventor for text similarity alignment and identifies different paraphrasing for the same text. The system is applied in evaluating machine translation by applying round-trip translation to generate different versions of same texts; there are 7 different languages used in this process. By utilizing the recall and precision of information retrieval, we used our solution to identify mis-translations. Finally, we run the system against different levels of plagiarism in a plagiarism corpus, to explore the ability of Dr Inventor of detecting plagiarism. The results represent significant progress towards achieving a fine-grained tool for text comparison.



MAHMOUD ABULEBDEH - STUDENT

BIOGRAPHY: Mahmoud Abulebdeh is a Software Engineer, born in Damascus, Syria. he holds a bachelor's degree in Information System Engineering from the SVU Damascus, Syria, currently he is finishing his first year as DESEM student at Maynooth University, Ireland. Mahmoud has a high technological, analytical, organizational and communication skills, gained from working in different countries and companies operating in different domains. Bbefore graduation, he worked as a web developer for "Teach Arabia", where he was involved in developing e-commerce and learning management systems, after graduation, he worked as a Software Engineer for "GWA", then he moved to Dubai, to work as a Project Manager and Digital Media Specialist for "Hattlan Media", after that he worked as web operation Engineer for coursetakers.ae. Then in 2015, he moved to Ireland to continue his Master studies at Maynooth University, where he had the privilege of being a Teaching Assistant, helping the students as lab demonstrator for the Software Engineering & Software Process module. Mahmoud is passionate about technology, he is willing to take on challenges and drive for results, motivated to continue to grow as a software engineer by learning new technologies and best practices especially in the following domains: Data Mining, Information Retrieval, Social Media Analytics, IoT, Cloud Computing and Virtual Reality.

PROJECT: SOFTWARE SYSTEM TO GATHER, FUSE AND INTEGRATE GEO-REFERENCED DATA FROM DIFFERENT SOCIAL MEDIA NEWTORKS Social media applications such as Foursquare, Instagram, Twitter, Flicker, etc. provides a powerful and extensive API (Application Programming Interfaces) to their data and information streams. It goes without saying that these data and information streams are of very significant volume and grow quickly in real time. Users of these applications often have geolocation enabled in the application and consequently their data and information has a geographical coordinated explicitly or implicitly embedded. This project aims to develop a piece of software which will be able to access these API and download/extract data and information from the API endpoints. This data and information will then be integrated into one dataset ideally in a database (such as SQLite or others). Download/extraction of data and information will be driven by location. So for example "download all data from these API for an area of 5KM around Maynooth University". This software when developed will be used in ongoing research into Volunteered Geographic Information and Place-based Social Media patterns.



ANDERS CANDASAMY - STUDENT

BIOGRAPHY: Anders is a student from the south of Norway. He is a developer at heart and loves learning and using new technology. He enjoys programming and is often found working on side-projects. His primary programming languages are Java, Javascript, Typescript and Scala.

PROJECT: MODEL-BASED DEVELOPMENT OF A HEMODIALYSIS MACHINE Development of Programmable Electronic Medical Systems (PEMS) is faced with several challenges. On one hand, we wish to be able to create PEMS in a cost effective manner. On the other hand, our PEMS must be proven for stringent safety requirements before being allowed to enter the market.

By using Event-B, we are able to create a system that can fullfill both of these requirements. We can create a model of the system in Event-B and use this model to check the safety requirements of the system. Once the model has been proven safe, it can be converted into machine code that can be used in the real system.



ANDREA BALOGH - STUDENT

BIOGRAPHY:

Andrea Balogh is a first year master student at Maynooth University, Ireland studying Dependable Software systems. She has been fascinated by compilers, or as she likes to call them: 'programming programs', even before starting her academic studies. Her other interests include languages, computational thinking, AI technologies and software engineering. Before coming to Ireland she finished her undergraduate studies in computer science at the University of Babes-Bolyai, Cluj-Napoca, Romania, where she wrote her undergraduate thesis in the area of code analysis and automated refactoring to design patterns. She gained work experience throughout the summers working for a small startup company in Cluj and then following a 3 months internship in Budapest, Hungary at Morgan Stanley. Next to her coding monkey life she also enjoys being an active participant of student unions, social events, or anywhere else where she can be part of the team spirit and the wolf pack. She speaks Hungarian, Romanian and English on a full professional level. In her spare time you can find her either on her bed reading a novel, or on a hiking trail living one out.

PROJECT: COMPUTATIONAL ANALYSIS OF GAMES THAT REQUIRE COMPUTATIONAL THINKING Some schools wish to teach computer science but starting it with teaching high level programming languages is not appropriate for many small children and young teenagers. It can be difficult to understand at that age the detailed functioning of computers and how code is processed and interpreted. Thus it is a real problem for computer scientists what topics and what approach to recommend for teaching at these levels. If this problem is not solved it is likely for schools to teach IT rather than computer science. So following the model of teaching other sciences it has been proposed to teach a lower level of details of programming by using a much simpler model of computation then a Turing equivalent programming language(C, Java, Python, etc). One of the most popular initiatives in this area is the Bebras challenge, whose goal is to promote Informatics and Computational Thinking among teachers and children of all age by organizing competitions and online challenges in more than 40 countries. In this project we design interactive games based on Bebras tasks and prove the computational power of these tasks to convince the reader that they are indeed a programming task using a special language for it.



KEYLOR CHAVES - STUDENT

BIOGRAPHY: Keylor is originally from Costa Rica, Central America. He holds a bachelor degree in Computer Science/Software Engineering from the Costa Rica Institute of Technology. For his bachelor final project, he completed an internship on machine learning and bioinformatics in the Intelligent Data Analysis Research Lab at the Czech Technical University in Prague, Czech Republic. He also completed a program on student leader-ship and culture and political studies of the United States at Saint Bonaventure Univer-sity, New York, as part of a scholarship from the United States Department of State. He has five years of experience as a software developer, architect and team leader. He has experience on mobile and full-stack web development for high-scale systems. He is currently completing the first year of the MSc in Dependable Software Systems pro-gram at Maynooth University with a scholarship from the Costa Rica Ministry of Tech-nology and works as a Senior Software Engineer Consultant for firms in Costa Rica and the United States. His is interested on web development, artificial intelligence and virtual reality. He enjoys traveling, learning new languages and cultures, and scuba diving. PROJECT: NAVWELL: EXPERIMENTAL SOFTWARE FOR STUDYING SPATIAL NAVIGATION ABILITIES IN HEALTHY PEOPLE AND PATIENTS The Morris water maze is an experiment that has been used by researchers in psychology and neuroscience to study the spatial abilities of animals, the influence of different brain regions in the process and to test models and treatments for diseases such as Alzheimer’s dementia. It consists on a pool, a goal platform and visual cues that the subjects use in order to escape the pool and reach the platform. Recently, it has been used on human participants but there is a lack of a soft-ware suite that allows researches to perform and manage these experiments. This project aims to take advantage of the advances in Virtual Reality, dependable software knowledge and create a system that would allow researches to design, perform and analyse experiments using a 3D virtual render of the Morris experiment on an Oculus Rift device. Our results show a dependable and high quality software suite where that the experiments can be designed, recreated in a virtual environment, and their results analysed in the same way that the original Morris water maze experiments. The resulting software suite constitutes a strong contribution to the scientific com-munity to advance on spatial navigation and brain behaviour research.



FADY ZAKI - STUDENT

BIOGRAPHY My name is Fady Zaki. I am about to complete my first year of the DESEM programme at the University of St Andrews. I graduated from the Engineering Faculty in Alexandria University in Egypt with a BSc degree in Computer Science in 2014. After finishing my undergraduate studies, I worked for 3 months as a software researcher in Alexandria University developing a mobile app that implements Augmented Reality (AR) concepts with indoor navigation solutions for interior design and real estate marketing. After that I worked for 8 months as a Junior Software Developer in POET Egypt LLC, an Egyptian subsidiary of POET GMBH in Germany. There, I worked on a web application project for registering purchase contracts and searching contract prices. I worked there until I got the acceptance for the DESEM programme. I applied for the DESEM programme because I am mainly interested in the applications of “hot” research areas, such as Software Architecture, in the industrial world. After finishing my postgraduate studies, I am hoping to pursue a career related to what I studied in the DESEM program especially in the field of Software Architecture.

PROJECT:

SPECIFYING SOFTWARE ARCHITECTURES IN JAVA Software architectures provide a high level model of a system capturing both functional and non-functional requirements. They can be used as the basis for development, maintenance, consistency and correctness checking and stakeholder communication. These uses require a precise specification of potentially multiple views of the architecture. Formal Architecture modeling can be used for checking consistency between the architecture specified and the implementation. However, these modeling techniques require expert knowledge, an additional notation and extra effort and time. The architecture specification and implementation are usually treated as separate artefacts, resulting in two different representations, the prescriptive architecture which captures the design decisions made prior to the system’s construction and the descriptive architecture which describes how the system has actually been built. These two architectures often end up being completely different from each other. This project aims to solve the above mentioned problems by integrating the architecture specification within the implementation. The main idea behind this project is to use Java constructs to capture the architectural constraints of the system. Integrating architecture within implementation has many benefits. Firstly, consistency checking between prescriptive and descriptive architecture is now easily automated. Secondly, the implementation can now serve as a documentation for the system’s architecture. The approach taken for solving this problem is through a Java library built on top of Bytecode analyzers. This library enables the architect to specify the system’s architecture and verify it against the actual implementation.



WEICHUNG SHAW - STUDENT

BIOGRAPHY My name is Weichung Shaw and I am from Taiwan, though I have spent some time pursuing education in the United States as well. I have a bachelor's degree in Electrical Engineering (with a second major in French) from the Johns Hopkins University. This has been my first year in the DESEM program - a year spent in the Université de Lorraine being versed in the profound art of formal methods. For hobbies I dabble in tennis, history, games and movies, literature both popular and classy, intoxication, opining, delusions of grandeur, siestas, etc.

PROJECT:

FORMAL VERIFICATION OF RAFT My project aims to outline a machine-checkable safety proof of the relatively new but quite popular consensus algorithm called Raft, created by Diego Ongaro and comprising his doctoral thesis. The logic used for the proof is the Temporal Logic of Actions (TLA), which comes with a software environment for composing and machine-checking proofs. In his thesis Ongaro has in fact provided a formal safety proof of his algorithm. That proof is, of course, stated in the English language. The point of machine-checkable proofs - which are stated instead in the language of a particular logic - is that English-language math proofs are not precise enough, sometimes leading to proofs that actually contain undetected errors. A machine-checkable proof, then, serves as a much more solid proof and a real guarantee. Furthermore, the formation of a machine-checkable proof naturally makes it desirable to consider a refinement approach, during which, as one goes from the concrete to the abstract, one discards the aspects of the algorithm which are not important to the proof and retain only the essential parts. This has the possible side benefit of making the proof argument more intuitive than even an English-language proof.



JUAN KARSTEN - STUDENT

BIOGRAPHY Juan Karsten is a DESEM student at University of st Andrews. He was born in Jakarta, Indonesia and studied in Universitas of Indonesia. During his undergraduate study, he became a teaching assistant in object oriented and cryptograhpy subject. In addition to that, he interned in PT Ruma as software engineer. After graduating from his undergraduate university, he worked as a software development engineer in GDP Venture, Indonesia for one year. He was responsible to develop web application and android mobile application. His main interest is artificial intelligence and machine learning. When he was in high school, he dreamt to study medicine. Thus, research in artificial intelligence and health will be fascinating for him. He is always a big dreamer. Joining dependable software system has made his dream come true: studying in international recognized university, learning new and exciting subjects, and living next to beach in St Andrews. The only disadvantage to be a big dreamer is he sleeps a lot. One thing he only hates from St Andrews is no spicy food here. His advice for people who will go to St Andrews: never ever believe “spicy” food in a restaurant here. The food is not spicy at all.

PROJECT:

ACCURATE AND AUTOMATIC LOCALISATION OF HUMAN VERTEBRAE FROM X-RAY COMPUTERD TOMOGRAPHY SCANS Human vertebrae localization is a vital step for several medical tasks, such as: diagnosing spine pathologies or abnormalities and planning surgery. However, manual vertebrae identification is a tedious process. Furthermore, a large amount of data to be localized makes manual localization approach implausible. Therefore, automatic and accurate localization is a necessity to minimize human effort and expedite the identification process. Automatic localization encounters difficult challenges due to spine abnormalities, e.g. kyphosis and scoliosis, and surgical implant. In addition to that, different computed topography(CT) machine produces different quality and intensity contrast of CT scan. CT scan result is also low resolution and lacks contextual information. Current state-of-the-art localization technique was developed by Microsoft research [1]. The technique is based on random classification forest. Nonetheless, classification forest needs dataset to be labelled densely. To overcome this problem, the technique transforms sparse annotation into dense probabilistic label. By utilizing these dense probabilistic labels, random forest is able to classify vertebrae with short range and local contextual features. Localization will be implemented in this project based on the technique developed by Microsoft Research Cambridge [1]. After implementing the current technique, it will be improved by using different features: entropy and binary pattern



DMITRY SMORODINNIKOW - STUDENT

BIOGRAPHY I’m a Masters student at National University of Ireland Maynooth studying Dependable Software systems. In summer 2011 I’ve graduated from Ural Federal University (Yekaterinburg, Russia) and was awarded a Bachelor degree (with Honors) in Information Technologies. In 2013 I’ve graduated with Honors Diploma in Masters Programme in Fundamental Informatics and Information Technologies of Ural Federal University (Yekaterinburg, Russia) where I obtained multidisciplinary theoretical and practical foundation to engage in designing and implementing Business Intelligence software and systems. In 2011-2015 I’ve worked as a .NET Software Engineer in the industrial company in Yekaterinburg. There I was involved in building Document Management System for the company in order to get rid of paperwork and transfer all document management to digital space. My interests are software architecture and solving algorithmic problems. Apart of programming, I love going in for sports, travelling and learning foreign languages.

PROJECT:

EVALUATION OF “SUBJECT-VERB-OBJECT” INFERENCES IN BLENDED SPACES. Conceptual Blending has been proposed as a general cognitive operation underlying our capacity for creativity. While conceptual blending is a powerful model for creativity, there are many challenges related to the computational application of blending. One of them is evaluation of inferences in Blended Spaces, which enables to distinguish between essentially meaningful inferences and useless inferences. In this project, by ‘meaningful’ and ‘useless’ we imply familiar and unfamiliar inferences in respect to common sense. To accomplish this, we rely on N-gram Language Model, smoothing methods and make use of such publicly available semantic networks as ConceptNet and IBM Watson Concept Insights. Also we work closely with Dr.Inventor project, which is a creativity support tool. Inferences that we work with in this project are of “Subject-Verb-Object” form. By assigning scores to “Subject-Verb-Object” inferences, we aim to predict how familiar they would be for human evaluators. Evaluation of inferences is carried out for Dr.Inventor project’s local corpus and for public web corpus as well. The results of our methods are then compared with human manual evaluation.



AKULINA LYAPUNOVA - STUDENT

BIOGRAPHY My name is Akulina Lyapunova and I am from Almaty, Kazakhstan. I am a first-year DESEM student, studying at the Maynooth University. In 2013 I graduated from the Bachelor’s course of Computing Systems and Software at Almaty University of Power Engineering and Telecommunication in Kazakhstan with the overall GPA 3,8 (of 4,0). During my studies, I have realized that I am interested in designing new programs and improving the existing code to make it more efficient. Moreover, I have a great interest in database development, so after graduation I have started my career in a bank as a database developer. Apart from my academic life, I have lots of interests. The main interest is learning foreign languages, as I have a desire to speak at least 7-8 foreign languages. During my undergraduate studies I started to learn Japanese and in 2 years of study, I got positive results in the Japanese-Language Proficiency Test (JLPT) at N3 level. Finally, in 2012 I got a chance to travel to Japan and take part in the program for Outstanding Students of Japan Foundation Japanese-Language Courses 2012, where I have made 35 friends from 20 countries all over the world.

PROJECT:

A PLUGIN FOR REFACTORING EVENT-B MODELS This thesis describes a plugin which was created for developers who intend to design their models using the Event-B language. Event-B is a formal language which uses mathematical techniques for system modelling and verification. The accuracy of the model is ensured by proof obligations. The main disadvantage of the Event-B is that it doesn’t have many well-developed modularization constructs and it is not easy to combine specifications in Event-B with those written in other formal languages. Developers can use the plugin described in this thesis if they want to make changes in the existing model such as element renaming or merging, without writing new elements from the scratch. Developers can use their knowledge of the institution theory and specification based operators to interact with the plugin. Software development requires from the developer not only an accurate and clear structuring of the system, as well as efficient tests for finding bugs, but also a strong mathematical proof. All these components allow software to be reliable and eliminate the possibility of system failure. The more complicated the system is, the more difficult it becomes to make sure that it works correctly and, in this context, mathematical proof can help to show the absence of bugs.



GUILLERMO REYES GUZMAN - STUDENT

BIOGRAPHY Guillermo Reyes was born in Aguascalientes, Mexico in 1986. He finished his bachelor degree in Mechatronics Engineering in the Universidad Panamericana in 2009. In 2008 he started working for Texas Instruments, in Aguascalientes, within the Education Technology branch of the company, as a Junior Software developer of the embedded team for the development of the operative system of the TI-NSPIRE. During the 2011-2012 he worked as a Senior Software Engineer in the Texas Instruments headquarters in Dallas, Texas. In January 2015 he was designed as Team Leader of the embedded team in Aguascalientes. From August of 2012 to June 2014 he also worked as lecturer imparting subjects like Data Structures, Algorithms and Artificial Intelligence at the Universidad Panamericana. Currently, Guillermo finished his first year in the Dependable Software System Master’s program in the Maynooth University.

PROJECT:

A SYSTEM FOR CAMERA TRACKING IN INDOOR ENVIRONMENTS Visual simultaneous localisation and mapping (visual SLAM) solutions allow a moving camera to simultaneously map it environment in 3D and determine its position and orientation (pose) relative to the map. While the most advanced visual SLAM systems require high-end graphics workstations, many use-cases only require mapping to be carried out once with the resulting map being used for localisation. The advantage here is that localising relative to a pre-existing map can be achieved using much less compute power. In this thesis we develop a system that takes advantage of this fact for resources constrained mobile platforms. Our aim is to permit tracking of mobile devices such as tablets or wearable cameras by building a system to compute the pose of a camera relative to a 3D feature map of an indoor environment. The map is first constructed by using Kintinuous \cite{nuimeprn5801} to compute a dense 3D map of the environment, including a graph of the associated camera poses. Next the video sequence is re-analysed to detect and track image keypoints in 3D across all the frames. Given an image from a camera in an unknown location, feature points are detected and matched to the map. The resulting correspondence are input to a 3D to 2D motion estimation technique to compute the camera's pose. The performance of the technique is quantitatively evaluated using multiple real-world sequences captured over a building scale environment.



RODRIGO RODRIGUES DE CARVALHO - STUDENT

BIOGRAPHY Rodrigo is from Belo Horizonte, Brazil, where he gained his BSc in Computer Science from Federal University of Minas Gerais in the year of 2008 and started his MSc in Dependable Software Systems in the year 2015. He is currently doing his first year at University of Saint Andrews and he is going to National University of Ireland in Maynooth for the second year. He also completed an MBA in the year of 2014 at Ibmec Business School, also in Belo Horizonte. He has been working as a Business Analyst for some software development companies in Brazil since the completed the undergraduate studies until he left to start the DESEM MSc.

PROJECT:

VISUALIZING AND ANALYZING SOFTWARE ARTEFACTS AND THEIR RELATIONSHIPS Artefact consistency management is a difficult problem in software engineering. Different artefacts, such as requirements specification, software architecture, design documents, source code and test cases are maintained separately, and evolve at different rates. Changes made to one artefact are not necessarily propagated to others that are also affected. This can lead to artefacts rapidly becoming inconsistent with one another and losing their value for development, maintenance and documentation purposes. Many factors contribute to making this problem difficult to solve: (1) artefacts are created using different tools, (2) there are many different artefacts with diverse representations and storage formats, (3) usually, there is no central storage for the relationships between them, (4) they are created at different times, (4) they are usually created and used by different people with different skill sets. Current approaches to address this problem only offer partial solutions since they are tool and artefact dependent. A holistic framework for artefact consistency is under construction at University of St Andrews Computer Science Department. The aim of this project is to create a visualization tool based on this framework which is neither tool nor artefact dependent. It will allow the visualization of dependencies within and between artefacts, and identify artefact elements that are impacted when changes are made to one representation of the system.



GÖKBERK KOÇAK - STUDENT

BIOGRAPHY: Gökberk Koçak was born in Ankara, Turkey in 2nd July 1991. He studied Computer Engineering in Galatasaray University, Istanbul. In 2nd year of his studies, he participated in IBM software academy competition with his colleagues and he won the 3rd price for the project of automated blood donation web-system. For international experience, he is accepted for student exchange program Erasmus to go to INSA Lyon, France. Besides computer engineering, he also studied industrial engineering as a double-major in Galatasaray University. During in his both studies, he worked on visual recognition and image processing on robotic systems and music listening trend analysis with artificial neural networks. After graduating from both degrees, for further international experience, he applied for Double Erasmus Mundus MSc in Dependable Software Systems and he is currently studying his first year in Maynooth University. His favourite programming language is Python and he has special interest in mathematical side of software development.

PROJECT: EVALUATION OF COOPERATIVE DEVICES IN HETEROGENEOUS SITUATIONS Given the immense catalogue of recorded music from the last 100 years that managing the material is becoming more and more difficult. However, at least nowadays having music in a digital format means that it can be distributed using easily. Furthermore, it is amenable to analysis using powerful AI algorithms which can assist on classification purposes. A significant problem is classification as this facilitates the building of libraries that can be accessible by users seeking musical material sharing common properties. Automatic classification needs to be carried out by a learning algorithm. Many different approaches are possible but in recent times the successful techniques are based around Artificial Neural Networks. This project is concerned with applying these techniques on audio classification. Specifically, it looks at classifying audio tracks based on a particular set of acoustic features that have been provided in the Million song data-set. Furthermore, the effects of loudness war can be also distinctive to analyse in addition the other features. The aim is to show that the classification technology can successfully predict the decade to which an audio track belongs to. Using a large database means that significant computational resources are required and another goal of the work is to use high performance computing technology to implement the algorithms. The results demonstrate that the method was successful in creating a good, computationally efficient, predictor.


ERASMUS MUNDUS

Erasmus Mundus MSc in Dependable Software Systems awarded ‘Best New Postgraduate Course of the Year 2013’

by GradIreland

Documents

ERASMUS MUNDUS - mscdesem.ie PRinter2.pdfErasmus Mundus Double MSc in Dependable Software Systems (DESEM). This is the first return of the DESEM summer school to Maynooth, since the