IFI Summer School 2015

The 2015 IFI Summer School is a week-long event for graduate students and reasearch assistants in informatics and related fields, where invited experts teach a number of different topics in day-long courses on a variety of topics in Computer Science.

The summer school will take place June 22-26, 2015 at the University of Zurich BIN (Department of Informatics, Binzmühlestrasse 14, 8050 Zürich). The courses will be held in two parallel sessions in room 2.A.01 and 2.A.10 from 9:00 - 17:00 (check-in starts at 8:45am) with coffee and lunch breaks. PLEASE NOTE: Dr. Angerer's course will take place in room 0.B.04!

All registered students are also invited to attend the summer school social event, which will take place on Wednesday, June 24th directly following the course. Details to follow.

The Summer School is open to doctoral students in computer science and related fields from the University of Zurich as well as other universities. Registration is free for IfI research assistants and ifi doctoral students. For all other students, fees are 90 CHF for the entire five-day summer school, or 20 CHF for individual courses. Attendance will be capped at 40 people per course.

Preference will be given to IfI doctoral students and research assistants and other participants will be admitted on a first-come-first-served basis.

Registration is now open and will close on Monday, 15th June 2015.

The fee will be paid on site in cash only at the Check-in desk outside the classrooms every day between 8:45 and 9:00am.

For UZH students, you can find the ECTS credit awarded by each course in the Overview above. For non-IfI students who would like to acquire cedits, you need to talk with the person who is in charge of credit transfering in your school first and find out if your school accepts/recognizes the ECTS credits awarded by IfI at UZH.

Instructor: Prof. Rachid Guerraoui

The hardware evolution is clearly towards multicore architectures and it might indeed well become the case that the only way for programmers to speed up their applications is by parallelizing them. Yet this is not easy and simple programming abstractions for synchronizing concurrent tasks are badly needed. The Transaction Memory (TM) paradigm is an appealing candidate abstraction.It is argued to be as easy to use as coarse-grained locking and as efficient as, hand-crafted, fine-grained locking.

Not surprisingly, a large body of work has been devoted to implementing the TM paradigm and exploring its ramifications. Very little work has however been dedicated to exploring the fundamental principles of the TM concept. Without these, it is simply impossible to establish the correctness of a TM algorithm, compare two algorithms, or determine whether certain performance limitations are inherent to the model or simply the result of an implementation artifcact. This tutorial will recall the TM paradigm, summarize the results of some of the fundamental work achieved so far on TM, and open new research perspectives for establishing TM correctness properties and verifying them.

Instructor: Prof. Geraldine Fitzpatrick

This course will consist of two parts. In the introductory part we will discuss research paradigms, "the set of common beliefs and agreements between scientists about how problems should be understood and addressed" (Kuhn 1962). Research paradigms are important because they set up what we think can be known (and how we frame our research questions and claims) and how it can be known (what methods we choose). This is important for your methodology chapter to frame and justify the overall approach of a research thesis. It is also important for understanding and valuing that there is a diversity of paradigms and respecting that different researchers can make different albeit equally legitimate choices.

The second part of the course will focus on qualitative research methods, most often used in a constructionist/interpretivist paradigm but also possible to apply across different paradigms. As applications become more complex and open-ended, the limitations of traditional software requirements techniques become evident and there is an increasing shift in both industry and academia to qualitative methods. This course will provide a practical introduction to qualitative user research methods for the purposes of technology design and evaluation. Topics will include: core qualitative methods of observation and interview; planning and conducting fieldwork; analysing qualitative data; moving from data to design; writing up qualitative research. Case studies from both research and industry will be used to illustrate the types of insights that can be gained. Participants will also have the opportunity to apply the methods in practical exercises.

Instructor: Prof. Andrea Rizzoli

Simulation models must be available at different levels of complexity, from the simplest, to efficiently syntethise control policies by means of optimisation algorithms, to the most complex, to deliver trustworthy “virtual worlds” for the exploration of alternative solutions and scenarios.

In this mini course we will examine how simulation and optimisation work hand in hand for the solution of complex planning and management problems.

Instructor: Prof. Kevin Crowston

Social network analysis as a field of study is growing rapidly and in popularity as an effective tool to study the ties or links between people, organizations and phenomena. It is both an approach and a tool to uncover and understand the connections that drive certain phenomenon involving interaction across a network. SNA techniques have been deployed to uncover and visualize hidden patterns in as diverse groups as academic communities to terrorists communities; as diverse phenomenon as correlating performance and creativity to who will be the next US president.

This class will provide a brief introduction to the concepts, methods and data analysis techniques of social network analysis. The course begins with a general introduction to the goals and perspectives of network analysis. This is followed by a practical discussion of network data, covering issues of collection, validity, visualization, and mathematical/computer representation. We then take up the methods of detection and description of structural network properties such as centrality, cohesion, subgroups, cores or roles. After the course participants will understand what it means to examine data in ’social networks way’.

Instructor: Dr. Christoph Angerer

Massively parallel computing is expanding rapidly: The extreme parallelism provided by GPUs are essential for many modern applications, from game graphics on the smallest smart phones to long-running scientific simulations on the largest super computers. In this hands-on lab, we will explore state-of-the-art programming techniques using CUDA and OpenACC paradigms and will dive into optimization, profiling, and debugging methods for modern GPU architectures. A basic understanding of C or C++ will be helpful but it is not required to have any prior experience with GPUs.

Instructors: Prof. Kunpeng Zhang

The development of web 2.0 and related technologies have led to an exponential increase in various types of user-generated content including textual and networked information. Finding meaningful nuggets of knowledge from such a big and diverse data has attracted a lot of attention. Hadoop and MapReduce as a well known distributed environment and computing framework have been widely and successfully deployed in many domains, particularly in the field of business, healthcare, and social media. Many scalable machine-learning algorithms such as clustering, association rule mining, recommender systems, topic modeling, and network analysis have been proposed and implemented in many open-source packages (e.g. Apache Mahout).

In this course, we plan to cover the following materials.

1)     Introduction to big data: basic concepts, characteristics, challenges, and applications.

2)     Distributed Hadoop system: architecture, installation and configuration.

3)     MapReduce programming: basic framework of MapReduce, advanced MapReduce framework (customized components). We have two examples to show to use MapReduce for solving real large problems.

4)     Algorithms: implementing three algorithms using MapReduce, including classic clustering (K-means), item-based collaborative filtering, and PageRank. In addition, we have hands-on exercises on (1) pseudo-distributed Hadoop installation and configuration; (2) MapReduce implementations: word count example and single-source shortest path finding.

Instructor: Prof. Margaret-Anne Storey

Computer Supported Collaborative Work (CSCW) concerns the design, use and evaluation of technologies that support teams, groups and communities. It is an interdisciplinary topic that addresses both the technical and social aspects of collaboration technology. In this one-day course, students will gain an understanding of the theoretical underpinnings of CSCW while exploring a wide range of existing collaboration and social technologies that are used in the domains of software engineering and education. The course will be highly collaborative, requiring students to work with each other through the use of social technologies and through small group discussions.

Instructor: Prof. Ben Lubin

This course will cover several key issues in the design and implementation of combinatorial markets, such as those used in the many recent billion dollar government spectrum auctions. We will first briefly review the motivation for such auctions and their design. The next part of the class will be focused on the particular problem of designing bidding languages for auctions that are both expressive and concise. We will then turn to the problem of pricing these auctions, with a specific focus on a surprising recent line of research that has established a connection between the pricing of combinatorial markets and kernel-based classifiers in machine learning (ML). Students will complete a hands-on exercise to help them rapidly understand the construction necessary for these ML techniques. We will then explain how to deploy such ML techniques in creating auctions of both theoretical and practical interest.

Instructor: Prof. Tobias Hossfeld

The purpose of the course is to develop an understanding how to evaluate the performance of Internet applications from a user-centric point of view. As an example, Internet video delivery is considered where users are interacting, e.g. by posting video links to friends, and watch the video accordingly. The focus is on the methodology how to measure Quality of Experience by means of crowdsourcing and how to analyze subjective results statistically sound. As a result, key QoE influence factors may be derived to formulate a QoE model.

The performance evaluation of Internet video delivery needs then to consider the network conditions as well as the video characteristics. This allows to quantify the user perceived quality. To this end, an optimal playout strategy is derived. The video buffer on application layer is modeled by a simple queueing system which is analyzed with common queuing theory methods. Together with the QoE model, a user-centric evaluation is possible.

The course will be organized as a whole-day event. Lectures with integrated exercises will be given to interactively implement the theory into practice for selected examples.

Contents of the course

·       Introduction to Performance Evaluation of Internet Applications

·       Crowdsourced Quality of Experience

·       Statistical Analysis of Crowdsourced Data

·       Modeling a video buffer with Queueing Systems

·       Optimization problem for QoE optimal video playout

·       Network analysis of video graphs

Instructor: Prof. Christof Fetzer

The dependability, in particular, the availability, integrity and confidentiality of systems can be impacted by both software as well as hardware issues. For example, hardware reliability issues already lead to diminishing performance returns when transitioning to smaller CMOS gate lengths. Bugs, both on hardware as well as software layers, can lead to dependability degradations.

In this course, I will introduce the approach and the resilience mechanisms that we are currently investigating in the context of cfaed (German excellence cluster with one focus on resilience) and SERECA (H2020 project on cloud security). The main objective of these projects are to reduce the cost of resilience. Therefore, the focus of this course will be on resilience mechanisms that are addressing both security as well as safety. Moreover, one of the goals of these projects is to be able to build safety critical systems on top of commodity hardware and software. In particular, we are investigating resilience mechanisms that permit us to offload some safety critical software components from cars to a cloud.

Prof. Rachid Guerraoui (EPFL)

Prof. Geraldine FItzpatrick (TU Wien)

Prof. Andrea Rizzoli (SUPSI)

Prof. Kevin Crowston (Syracuse University)

Dr. Christoph Angerer (NVIDIA)

Prof. Kunpeng Zhang (University of Illinois)

Prof. Margaret-Anne Storey (University of Victoria)

Prof. Ben Lubin (Boston University School of Management)

Prof. Tobias Hossfeld (University of Duisburg-Essen)

Prof. Christof Fetzer (TU Dresden)