Distributed Systems Seminar -- Spring 2016
Schedule for Spring 2016: TUE 14:15 Liivi 2-512
Organization and requirements
16.02.2016: First meeting, topics introduction
23.02.2016: No meeting
01.03.2016: Topic choice, consultation on how to proceed
08.03.2016: Meeting for Research Plan and initial Literature Review presentations
15.03.2016: (No meeting, Individual work, Meetings with supervisors)
22.03.2016: (No meeting, Individual work, Meetings with supervisors)
29.03.2016: No meeting, Do the tests and experiments, Start writing the final report, provide the supervisors with an access to the code as well as to the report draft
05.04.2016: Meeting: Intermediate presentations / Technical Demo, First draft of the final report ready for Peer-review
12.04.2016 (No meeting, writing the report)
19.04.2016 (No meeting, writing the report)
26.04.2016: Last day to share the reports between the reviewers
03.05.2016: Last day to the reviewers to give their feedback10.05.2016: The day of the final presentation and handing out the final report.
Seminar papers:
- Siim Plangi: Overview of Backend as a Service
- Jaan Janno: GPGPU n-body simulation for 2-D games in Java
- Joonas Lõmps: Recognising Hiragana Characters Using Optical Character Recognition (First Place Winner in the Student Project Contest June 2016)
- Joosep Kibal: Real-time Hidden markov model based map-matching
- Mati Kärner: Trajectory prediction in cellular networks using Discrete-time Markov Chains
- Andrii Rozumnii: Implementation of exact-pattern matching algorithms using OpenCL and comparison with basic version code
- Asmar Hasanova - Difference of Distributed Systems Course at University of Tartu and Other European Universities
- Elis Kõivumägi: Semantically enriching mobile subscriber's trajectories
- Siim-Toomas Marran: Microservices log gathering, processing and storing
- Deliverables:
- Written results report (in LaTex IEEE technical report format)
- Research plan (has to be provided in one week after topic selection)
- Weekly report in form of blog on our web page (the access will be granted)
- Research results: The code in our SVN repository
- Research results: The collected related work publication list on our web page
- Research results: The technical report LaTeX sources in our SVN
- Intermediate presentations / Technical Demo (in the middle of the term)
- Peer-review of the report drafts between the seminar members
- Final peer-review
- Final presentation / technical demo
You can see the final reports online at: Previous: Distributed Systems Seminar (Fall 2015)
Seminar supervisors: Prof. Eero Vainikko, Dr Amnir Hadachi, Dr Benson Muite, Artjom Lind, Oleg Batrašev
Topic Areas (Concerned persons):
- Parallel Scientific Applications and Concurrent Computing (Eero Vainikko, Oleg Batrashev, Benson Muite)
- P2P Computing: F2F Platform, F2F Applications (Artjom Lind)
- Applied Computer Vision (Artjom Lind)
- Parallel Machine learning algorithms (Artjom Lind, Oleg Batrashev, Amnir Hadachi, Benson Muite)
- Exploratory search (Dimitri Danilov)
- Geographic information systems (GIS) Related topics (Amnir Hadachi)
LaTeX Resources
Literature Search Resources
List of possible topics
Possible themes with some suggested materials to start with:
1. Parallel Scientific Applications and Concurrent Computing (eero at ut.ee)
-
deal.II — an open source finite element library
- Ways for parallelising serial python and numpy codes
- Multi-GPGPU Computing (with focus on speeding up weather forecast calculations) -- GPU memory-mapping restructuring for better overlap between data movement and calculations
- Intel Xeon Phi architecture for speeding up calculations
- Computing for multicore systems (olegus at ut.ee)
- Charm++
- Intel Thread Building Blocks (TBB) - task parallelism in C++
- any other library for multicore systems
- Libraries and languages for Concurrent programming (olegus at ut.ee)
- SPA: Angular JS, ... (two-way synchronization)
- Facebook React.js
- Clojure(script): re-frame, Reagent
- any other language for concurrent programming
- Languages for scientific computing (all Partitioned Global Space languages)
- Chapel
- Coarray Fortran -- http://www.co-array.org/, http://caf.rice.edu/, intel
- Unified Parallel C -- http://upc.gwu.edu/, http://upc.lbl.gov/, http://www.gccupc.org/
- Trends in algorithm development for shared memory architectures, lock-free datastructures etc.
2. Parallel Scientific Applications and Concurrent Computing (benson punkt muite at ut punkt ee)
- Scripting and high level language interfaces (eg Java and Python) to
the Message Passing Interface and/or to OpenCL
- Information security on distributed systems
- Automatic performance tuning of numerical algorithms (eg. Spiral, Atlas or FFTW )
- In-situ visualization of solutions to differential equations on GPUs/Xeon Phi and/or multi cpu computers
- Efficient time integration schemes, geometric integration schemes
- Particle methods for differential equations
- Parallel integral equation evaluation for volume rendering
- Low power efficient heterogeneous computing; comparison of tuning for power efficiency and tuning for speed
- Good parallel software development practices
- Testing and improvement of FortranCL
- Financially sustainable open source parallel software development models and practices
- Testing and improvement of Seedme for distributed computing
- Testing MPPA accelerators such as Kalray and Parallela
- Testing FPGAs from Altera and Xilinix
- Parallel Graph Algorithms, for example GraphBLAS
3. Peer-to-Peer (P2P) Computing Architectures (Artjom Lind)
Covering the topics related to distributed computing in peer-to-peer networks. Here I will focus you on own lab framework, however you can propose different topic (some existing framework with similar features or your own design).
- Individual topic -> Contact me!
- Friend-to-Friend (F2F) Platform
- Friend-to-Friend (F2F) Applications
4. Applied Computer Vision (CV) (Artjom Lind)
Mostly the topics related to the application of latest results in CV. In this area we mostly use OpenCV library, which is recommended but not obligatory. The several topics we can focus on:
- Structure from motion
- Object detection/classification
- Object tracking
- Optical Character Recognition (OCR)
- Augmented Reality
5. Parallel Machine learning algorithms (Artjom Lind, Oleg Batrashev, Amnir Hadachi, Benson Muite)
- Machine digitization and translation for Estonian and non-latin scripts such as Arabic / Cyrillic / Chinese / Farsi / Hebrew / Hindi / Japanese / Korean
- Character recognition algorithms
- Evaluating Petuum
6. Exploratory search (Dimitri Danilov)
- Collaborative search
- Search Patterns
- New Development in Search Engines
- The Vision of Ted Nelson (the inventor of the internet?)
- Xanadu (and undanax)
- Graph Based Information Storage
- New Search Interfaces in Mozilla (practical and theoretical topics available)
- Machine learning based topic modeling in text documents (using the program Mallet).
- More topics on demand
7. Modeling and analyzing semantic trajectories (Amnir Hadachi)
|
|
![]() |
8. Mobility data modelling (Toivo Vajakas)
Fast data structure for trajectory data (further development of existing code)
o Currently existing code for direct read of 1 individual trajectory
o Add support for batch full-scan and batch subset.
o Add support for indexing by time and space – to avoid (with high probability) the analysis of trajectories that do not intersect with time-space volume of given query.
Exploratory data analysis on mobile positioning data, using results of Jilles Vreeken group (JV was a speaker on ESSCASS summer school)
o Separation of time of each person as „currently in routine“ and „non-typical behavior (tourist mode)“.
o Describe the data, ie the behavior of radio network combined with behavior of people -- pattern mining based on information theory.
Clustering
o Clustering of (relatively small) directed graphs, each edge and vertex has also attributes. (graphs are HMM of humans in timespace „states“ like work and home, after removing absolute location and travel direction info (but keeping relative travel distance info)
o Clurstering of vertices in directed graphs (states of HMM in previous entry). Graphs are HMM of humans in timespace „states“ like work and home, after removing absolute location and travel direction info (but keeping relative travel distance info). Clustering of states gives something like „here many people spend there night“, „many people come here for work“, ...
Combination of traffic simuluation package and data from mobile positioning data, to get local traffic density estimates.
9. Mobility data modeling and representation (Amnir Hadachi)
|
![]() |
|