Skip to end of metadata
Go to start of metadata


ATTENTION! Update 25-May-2021

The CS Colloquium 2020-21 is now over.



For the events of 2020-21, the Colloquium will take place online, via live and recorded presentations. The will be no physical meeting.

Links to the online meetings will be posted on Moodle a couple of days before the event.

Contact

For any issues or questions, contact the main instructor, Michael Mathioudakis (michael.mathioudakis@helsinki.fi).
In the exceptional case when a matter is urgent and the main instructor is on sick-leave, contact Veli Mäkinen (veli.makinen@helsinki.fi).

Information for MSc thesis presenters

Who presents and when

  • Students in the CS Master's Programme (2017 onwards) should give one presentation on their Master's thesis! This is a mandatory part of the 30 cr thesis project.
  • Students in the CS study programme (prior 2017) are also encouraged to give a presentation on their 40 cr Master's thesis; in that case, 1 cr is given just from the presentation without other obligations.
  • Negotiate with your supervisor a suitable time (i.e., CS Colloquium event).
    • A good time to present is when there are some results to show, but the thesis is not yet fully written.
    • The purpose of the presentation is to give a final boost to finalize the project.
    • Feedback from the audience can give new ideas.
  • Once you have chosen when to present, register through this e-lomake.
    • Deadline: 1 week before the event, before 12:00 (noon).
      For example, if the event takes place on Friday, then the deadline to register is by noon of the previous Friday.

Live and recorded presentations

  • For each event, there will usually be some presentations shown during the event and some recorded ones made available online.
    During the event, presentations can be presented live or be played from a recorded video.
    All presenters must attend the event, regardless of whether they have a live or recorded presentation.
    Exceptions can be made only for medical reasons, with a doctor's note.
  • After you register, the instructor will notify you whether your presentation will be presented during the event.
    You can choose whether to present it live or via a recorded video.
    • The presentations shown during the event will be determined based on the available time of each event (e.g., there may be a keynote or faculty presentations during the event). Presentations are selected so that (a) priority is given to students who register early, and (b) different study tracks are covered.
  • Read the Instructions for presentations.
  • The schedule of each event will be fixed and moved to the Moodle one week before the event.
    If you do not have access to moodle, contact the instructor.

Information for attendees

See the Moodle of the course.

Topics for Master's Thesis

Archive 2017-2018 – Archive 2018-2019  – Archive 2019-2020


Schedule 2020-2021

Fri, 11/09/2020, 2:15 pm - 4:00 pm

The event takes place on Zoom.

Schedule

2:15 pm : Introduction to the CS Colloquium by Michael Mathioudakis

2:30 pm : Keynote by Prof. Aapo Hyvärinen, Computer Science, University of Helsinki.

Title: Unsupervised Deep Learning

Abstract: Unsupervised learning, in particular using neural networks or deep learning, is one of the most important problems in machine learning. Unsupervised learning actually groups together a rather disparate set of learning methods with different goals, such as latent variable estimation, density estimation, generation of new data points, causal discovery, feature extraction, etc. In this talk, I give a short overview of my recent work on several of these topics.

Bio: Aapo Hyvarinen studied undergraduate mathematics at the universities of Helsinki (Finland), Vienna (Austria), and Paris (France), and obtained a Ph.D. degree in Information Science at the Helsinki University of Technology in 1997. After post-doctoral work at the Helsinki University of Technology, he moved to the University of Helsinki in 2003, where he was appointed Professor in 2008, at the Department of Computer Science. From 2016 to 2019, he was Professor at the Gatsby Computational Neuroscience Unit, University College London, UK. Aapo Hyvarinen is the main author of the books "Independent Component Analysis" (2001) and "Natural Image Statistics" (2009), and author or coauthor of more than 200 scientific articles. He is Action Editor at the Journal of Machine Learning Research and Neural Computation, and has worked as Area Chair at ICML, ICLR, AISTATS, ACML and NIPS. His current work concentrates on unsupervised machine learning and its applications to neuroscience.

3:00 pm : MSc Thesis Presentation

3:20 pm : Other activities/announcements


Fri, 09/10/2020, 2:15 pm - 4:00 pm

The event takes place on Zoom.

2:15 pm: Introduction by Michael Mathioudakis

2:20 pm: Keynote by Eric Malmi, PhD, Google Zurich.

Title: Natural Language Generation via Text-Editing Models

Abstract: Natural language generation (NLG) has a wide range of applications, from building digital assistants to automatic news generation and grammatical error correction. However, deploying NLG methods in the wild is challenging for many reasons, including the tendency of the traditional sequence-to-sequence methods to “hallucinate” non-factual information and the slow inference speed of these methods. To tackle these limitations, we have developed a set of text-editing methods, which address a class of NLG tasks where a source text needs to be rewritten in a given target style. This talk gives an overview of these methods, discussing their trade-offs as well as the challenges related to building real-world NLG applications.

Bio: Eric Malmi is a software engineer at Google, Zürich. His research focuses on developing machine-learning methods for natural language generation. He holds a PhD (2018) in computer science from Aalto University, and his research has been widely featured in the media, including Washington Post, Bloomberg TV, and The Wall Street Journal.

3:00 pm: MSc Thesis presentations

  • Joonas Sarapalo, Hit counting in Amazon Web Services. Supervisor: Jussi Kangasharju.
  • Peter Porttinen, Chaining with maximal exact matches. Supervisors:   Veli Mäkinen, Jarno Alanko.
    Article: https://arxiv.org/abs/2001.06864
    Comment:   The chaining algorithm described in the paper is implemented and is central to the result. Hint: Finding all maximal exact matches between two strings is also a significant part of the thesis; possible tailored assignment could involve suggesting new ideas for finding the matches or uses for the chaining outside the alignment described.

3:40 pm: Conclusion by Michael Mathioudakis



Software

  • Jussi Liukkonen, Sovelluksien päivitysprosessien automatisointi pilvipalvelussa. Supervisor:  Antti-Pekka Tuovinen.
    Article: https://ieeexplore.ieee.org/document/8776985.
    Comment: Yleiskuva jatkuvasta integraatiosta ja toimittamisesta, pilvipalveluista sekä sovellusten pyörittämisestä konteissa.

Algorithms

  • LIVE: Peter Porttinen, Chaining with maximal exact matches. Supervisors:   Veli Mäkinen, Jarno Alanko.
    Article: https://arxiv.org/abs/2001.06864
    Comment:   The chaining algorithm described in the paper is implemented and is central to the result. Hint: Finding all maximal exact matches between two strings is also a significant part of the thesis; possible tailored assignment could involve suggesting new ideas for finding the matches or uses for the chaining outside the alignment described.
  • Arpita Seth, A Comparative Study Between Local and Global Approaches for Hierarchical Classification. Supervisor: Indre Žliobaitė.
    Article: https://www.cs.kent.ac.uk/people/staff/aaf/pub_papers.dir/IEEE-ICDM-2009-Silla.pdf
    Comment:   The master thesis has two parts - first comparing local and global approaches of hierarchical classification and in next part deals with different approaches of global hierarchical classification on data streams. Link for data stream : https://arxiv.org/abs/1010.4784

Networks

  • LIVE: Joonas Sarapalo, Hit counting in Amazon Web Services. Supervisor: Jussi Kangasharju.

Fri, 13/11/2020, 2:15 pm - 4:00 pm

Live Industry Talk

  • Frederick Ayala, Senior Data Scientist at Rovio. Topic: Supporting mobile games growth using data science.

    Abstract:
    Developing mobile free-to-play (F2P) games is about constant improvements to the game and its operations. To know if the game is improving or not, game developers collect large amounts of data from different sources. In this talk, we present an overview of the data that mobile F2P game developers collect and show some of the work that the data science team at Rovio does to help the games grow. We also briefly describe the challenges that iOS14 brings to the mobile industry.

    Short Bio:
    Frederick Ayala is Sr Data Scientist and Team Manager at Rovio. He has a Ph.D. in Information Systems and Technology from ELTE University, and has experience as data scientist and data engineer in industries such as mobile gaming, telecommunication, and retail.

Live Session about Doctoral studies at the CS Department

  • Prof. Tomi Männisto (director of the CS doctoral programme), Pirjo Moen (Research coordinator), Sini Karppinen (Doctoral admissions coordinator).

Live MSc thesis presentation

  • Noora Mäkelä. Topic: Structure and Parameter Learning of Sum-Product Networks. Supervisor: Mikko Koivisto.

Software

  • Risto Mikkola, "Tulkin sulauttaminen monisäikeiseen ohjelmaan". Supervisor: Tommi Mikkonen.
    Article: https://doi.org/10.17771/PUCRio.DImcc.24342
    Comment: Paperi on lähellä omaa työtäni ja kertoo oleelliset asiat. Paperissa monisäikeinen isäntäohjelma sulauttaa Lua-kielen tulkin ja käyttää sitä eri säikeistään suorittaakseen kyseisen kielen ohjelmia rinnakkain. Tehtävät voisivat liittyä mahdollisten ongelmien pohtimiseen isäntäohjelman ja sulautetun tulkin välillä, kun isäntäohjelma on monisäikeinen. Esimerkiksi voidaan pohtia sulautuksen toteutuksen vaikutusta tehokkuuteen, säikeidenvälisen tiedonvälityksen ongelmia, sulautetun kielen tulkin ja isäntäohjelman tiedonvälityksen ongelmia tai erilaisten tietotyyppien esittämisen ongelmia kielten välillä. Voidaan myös pohtia, minkälaiset kielet soveltuisivat hyvin monisäikeiseen sulauttamiseen, minkälainen arkkitehtuuri olisi sopiva erilaisissa tilanteissa tai olisiko jonkinlaista yleistä ratkaisua sulauttamisen arkkitehtuurille.
  • Tuomas Siipola, "Päätöksenteko ohjelmointikielten jatkokehityksessä: tapaustutkimus Rustin yhteisöllisestä kehityksestä". Supervisors: Tommi Mikkonen.
    Article: https://livablesoftware.com/governance-strategies-programming-languages/
    Comment:   Katsaus ohjelmointikielten kehittäjäyhteisöjen rakenteeseen ja niissä käytettyihin päätöksentekomalleihin.
  • Topi Törnroos, "APM tool requirements study for Veikkaus". Supervisors:   Jukka Nurminen (HY), Tuomas Lempiäinen (Veikkaus).
    Article: https://en.wikipedia.org/wiki/Application_performance_management
    Comment:   The Finnish government-owned agency Veikkaus Oy is overhauling its network infrastructure, and as a part of the project, a requirements study was performed regarding the current state of the art Application Performance Management tools and Veikkaus's own needs regarding them. I performed a requirements analysis and a comparison of available products for my thesis.
  • Suvi Sipilä, "Puhdas ja laadukas koodi luokka- ja funktiotasolla",  Supervisor: Antti-Pekka Tuovinen.
    Article: https://ieeexplore.ieee.org/document/8587297
    Comment:   Artikkeli käsittelee hyvin niitä tekijöitä, jotka tekevät koodista puhdasta ja laadukasta. Artikkelissa on samantapainen tutkimuskysymys kuin omassa työssäni: "Miten puhdas ja laadukas koodi toteutuu käytännön työssä ohjelmistokehittäjien näkökulmasta?".
  • Satu Kärkkäinen, "Museoiden digitaalinen yleisötyö". Supervisor: Tomi Männistö.
    Article:   https://julkaisut.valtioneuvosto.fi/bitstream/handle/10024/161674/OKM_2019_25_Kulttuurilaitosten_digitaalinen_yleisosuhde.pdf?sequence=1&isAllowed=y
    Comment:   Museot hyödyntävät digitaalisia keinoja ja kanavia yleisötyössään monin eri tavoin. Niihin liittyy erilaisia haasteita ja huomioitavia seikkoja. Kuinka keinot ja kanavat valitaan ja miten yhteistyötä eri toimijoihin voidaan hyödyntää.

Algorithms

  •  Juhana Laurinharju, Talk title:   Optimizing Phylogenetic Supertrees with MaxSAT. Supervisors:   Matti Järvisalo, Emilia Oikarinen.
    Article:  https://doi.org/10.1017/S1471068415000265
    Comment:   Solves the problem using a higher-level optimization technique.

Networks

  • Mikko Rinta-Homi, "Controlling heating, ventilation and air conditioning with IoT to save energy". Supervisors:  Petteri Nurmi, Agustin Zuniga.
    Article:   https://www.sciencedirect.com/science/article/abs/pii/S0378778818320140
    Comment:   This material covers well both background of the topic and the use of low-resolution thermal imaging cameras in tracking, which are also used in the research part of the thesis.

Fri, 15/01/2021 2:15 pm - 4:00 pm

Live Keynote

Al­ex­an­dru Tomescu, Associate Professor, University of Helsinki. "Safe solutions: A link between algorithms and bioinformatics".

Live Discussion

Doctoral positions in Computer Science
Prof. Jukka Nurminen, Prof. Keijo Heljanko, Prof. Luigi Acerbi, Prof. Kai Puolamaki, Dr. Tuukka Ruotsalo

MSc Thesis

Fri, 05/03/2021 2:15 pm - 4:00 pm

MSc Thesis

Live activity

  • Explore CS Faculty research.


Fri, 09/04/2021 2:15 pm - 4:00 pm

Software

Algorithms


Tue, 18/05/2021, 1:00pm - 4:00pm

Zoom Linkhttps://helsinki.zoom.us/j/69007490293?pwd=MFc4cGREZXBZaHRkOWxXRFJRVFBXUT09
Meeting ID: 690 0749 0293,  Passcode: 570597

SCHEDULE (Final)

Part 0: 13:00 Introduction (Michael Mathioudakis)

  • 13:05 Mikko Koivisto, Director of the CS Master's program

Part 1: 13:15 - 14:15 :: Parallel Sessions A1, B1, C1, D1

Part 2: 14:30 - 15:30 :: Parallel Sessions A2, B2, C2


Sessions

# Session A1 (Chair: Jukka Nurminen)

Victor Bankowski: Dynamic Linking in WebAssembly

Paulius Daubaris: Towards Adaptive WebAssembly Applications

Aleksi Huotala: Benefits and Challenges of Isomorphism in Single-Page Applications: A Case Study and Review of Gray Literature

# Session A2 (Chair: Tommi Mikkonen)

John Lång: Modelling a Distributed Data Acquisition System (30' talk)

Santeri Juslenius: WebSocket vs WebRTC in the stream overlays of the Streamr Network

-------

# Session B1 (Chair: Laura Ruotsalainen)

Niclas Joswig: Artificial Intelligence for Industry Vision.

Otto Hyytiälä: Real-time satellite data processing platform architecture

Matti Leinonen: Monocular 3D Object Detection And Tracking in Industrial Settings

# Session B2 (Chair: Antti-Pekka Tuovinen)

Juuso Heikkinen. A Benefit Analysis of Utilizing Test Automation in a Telecommunications Company From a Business Perspective

Matias Wargelin. Building an evaluation environment for musical pattern discovery algorithms

Yu Zhang: Security Issues in IoT Networks

-------

# Session C1 (Chair: Michael Mathioudakis)

Leevi Lehtonen: Quantum Boltzmann machines.

Ilmo Salmenpera: Training Quantum Restricted Boltzmann Machines Using Dropout Method.

Juha Harviainen: Approximating the Permanent of a Matrix with Deep Rejection Sampling.

# Session C2 (Chair: Michael Mathioudakis)

Verner Saaristo: Military Command and Control systems and Simulators interoperability

Damian Kone: High Availability Systems

Mikko Pyykkö: Sharing DevOps knowledge in small organizations

Riku Alho: Systematic Literature Review on the Modularity of Modular Neural Networks

-------

# Session D1 (Chair: Petteri Nurmi) UPDATE 17.05: The order of presentations changed.

Jarmo Kallio: Making ERP system UI better for novice and expert users

Pinja Talosela: Ohjelmisto lääkinnällisenä laitteena

Hyeongju Lee: Scalable and High Available Kubernetes Cluster in Edges for IoT Applications



Thesis information

Software

  • Presenter name: Ilmo Salmenpera.
    Talk title: Training Quantum Restricted Boltzmann Machines Using Dropout Method.
    Supervisors: Jukka Nurminen
    Article: https://arxiv.org/abs/1510.06356
    Comment: Article describes the overall architecture for how quantum annealing can be used to train Deep Belief Networks, the general benefits of the approach and the key problems in the area.

  • Presenter name: Verner Saaristo.
    Talk title: Military Command and Control systems and Simulators interoperability.
    Supervisors: Tommi Mikkonen, Juha Vihavainen
    Article: https://science-sciencemag-org.libproxy.helsinki.fi/content/sci/323/5910/72.full.pdf
    Comment: Fletcher, J. (2009). ?Education and training technology in the military?. science323.5910,s. 72?75. The content of my thesis isn't focused on simulation per se, but the topic of C2SIM requires an understanding of the benefit of both simulators and command and control systems. To my mind, understanding the benefits of simulators is more accessible. If such considerations are unfounded the following article: https://netlab.gmu.edu/pubs/12S-SIW-045-20Feb-submit.pdf Which details the usage of two different standards to provide interoperability between C2-systems and simulators. These standards eventually converged to the C2SIM-standard which is central to my thesis.

  • Presenter name: Victor Bankowski.
    Talk title: Dynamic Linking in WebAssembly.
    Supervisors: Tommi Mikkonen, Niko Mäkitalo
    Article: https://www.usenix.org/system/files/atc20-shillaker.pdf
    Comment: Faasm is a Wasm runtime meant for serveless computing. What's particularily relevant about it in relation to my thesis is that they have implemented a Posix like dynamic linking interface which ended up influencing our dynamic linking implementation. The paper mentions the interface relatively briefly and so most of the paper is a bit tangential to my topic.
    However it does illustrate an interesting use case for wasm that also includes dynamic linking and so it provides insight into why wasm and dynamic linking are an interesting research topic.

  • Presenter name: Otto Hyytiälä
    Talk title: Real-time satellite data processing platform architecture
    Supervisors: Laura Ruotsalainen & Lauri Häme
    Article: https://www.mdpi.com/2072-4292/12/8/1253
    Comment: An Overview of Platforms for Big Earth Observation Data Management and Analysis
  • Presenter name: Aleksi Huotala
    Talk title: Benefits and Challenges of Isomorphism in Single-Page Applications: A Case Study and Review of Gray Literature
    Supervisors: Dr. M.Luukkainen, Prof. T.Mikkonen
    Article: https://www.manning.com/books/single-page-web-applications
    Comment: This book is a great introduction to the Single-Page Application (SPA) architecture. With a little Googling you can find the book's PDF (not linked here for obvious reasons).
    This second article gives a good introduction to Isomorphic web application architecture:
    https://medium.com/@ElyseKoGo/an-introduction-to-isomorphic-web-application-architecture-a8c81c42f59

  • Presenter name: Juuso Heikkinen
    Talk title: A Benefit Analysis of Utilizing Test Automation in a Telecommunications Company From a Business Perspective
    Supervisors: Antti-Pekka Tuovinen
    Article: https://www.agileconnection.com/sites/default/files/article/file/2014/Cost-Benefit%20Analysis%20of%20Test%20Automation.pdf
    Comment: The article describes how a simplified cost-benefit analysis for test automation can be made. Based on it, a benefit analysis for the target company was formed. As the study revolved heavily around testing and test automation in a company utilizing the principles of Scaled Agile Framework (SAFe), the following article gives a little more insight on SAFe and the target company Telia: https://www.scaledagileframework.com/case-study-telia-finland/
  • Presenter name: Mikko Pyykkö
    Talk title: Sharing DevOps knowledge in small organizations
    Supervisors: Tomi Männistö, Petri Ihantola
    Article: https://www.researchgate.net/publication/335272884_Closing_the_IT_Development-Operations_Gap_The_DevOps_Knowledge_Sharing_Framework
    Comment: The background material describes some essential concepts in DevOps and knowledge sharing.
  • Presenter name: John Lång
    Talk title: Model Checking a Distributed Data Acquisition System
    Supervisors: Keijo Heljanko
    Article: https://link.springer.com/chapter/10.1007/978-3-319-10575-8_1
    Comment: This chapter of the handbook gives a short introduction to model checking.
  • Presenter name: Riku Alho
    Talk title: Systematic Literature Review on the Modularity of Modular Neural Networks
    Supervisors: Jukka Nurminen
    Article: https://arxiv.org/pdf/1902.09240.pdf
    Comment: This article tests the concept of modularity with a modular neural network experiment.
  • Presenter name: Jarmo Kallio
    Talk title: Making ERP system UI better for novice and expert users
    Supervisors: Mikko Kytö
    Article: https://dl.acm.org/doi/abs/10.1145/1632149.1632162
    Comment: Aim in my thesis work was to investigate usability of an ERP system used by many middle sized ferry companies. The current understanding had been that the system was easy and efficient for long time users but hard for new people to learn. So I conducted interviews and asked new and old users to do some tasks with the system to see what kind of problems could be found. The aim after that was to give possible recommendations what things should be kept in mind when people start to implement new version of this system.

  • Presenter name: Pinja Talosela 
    Talk title: Ohjelmisto lääkinnällisenä laitteena
    Supervisors: Tommi Mikkonen
    Aerticle: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7381013/
    Comment: Johdanto ohjelmistokehittäjille lääkinnällisiin ohjelmistoihin liittyvästä sääntelystä ja yleiskuva myös uudesta EU:n lääkinnällisten laitteiden asetuksesta.

Algorithms

Networks

  • Presenter name: Paulius Daubaris
    Talk title: Towards Adaptive WebAssembly Applications
    Supervisors: Prof. Tommi Mikkonen, Dr. Niko Mäkitalo
    Article: https://dl.acm.org/doi/10.1145/1985484.1985496
    Comment: The article describes a similar system and provides enough information to understand the approach presented in the thesis.
  • Presenter name: Damian Kone
    Talk title: High Availability Systems
    Supervisors: Niemi, P. Valtteri
    Article: https://ebookcentral-proquest-com.libproxy.helsinki.fi/lib/helsinki-ebooks/detail.action?pq-origsite=primo&docID=875747
    Comment: This book is a basis for an advanced course on high service availability systems in graduate studies or in continuous education.
  • Presenter name: Santeri Juslenius
    Talk title: WebSocket vs WebRTC in the stream overlays of the Streamr Network
    Supervisors: Samu Varjonen, Eric Andrews
    Article: https://ieeexplore.ieee.org/document/7889934
    Comment: Performance comparisons for TCP and SCTP, the underlying transport layer protocols for WebSocket and WebRTC.
  • Presenter name: Yu Zhang
    Talk title: An Overview: Security Issue in IoT Network
    Supervisors: Niemi, P Valtteri
    Article: https://ieeexplore.ieee.org/document/7412116
    Comment: Internet of things (IoT) security: Current status, challenges and prospective measures
  • Presenter name: Hyeongju Lee
    Talk title: Scalable and High Available Kubernetes Cluster in Edges for IoT Applications
    Supervisors: Naser Hossein Motlagh, Petteri Nurmi
    Article: https://more.suse.com/rs/937-DCH-261/images/ARM-White-Paper-V3.pdf
    Comment: Though I did not refer anything from this article, most of background knowledge (cloud, edge, iot, container, Kubernetes, scalability, HA) are explained based on K3s. I also used K3s distribution for the implementations in my thesis.
  • No labels