Group Projects: ScaDaMaLe WASP Instance 2022-2023

Edited by Oskar Åsbrink and Raazesh Sainudiin.

Peer-reviewed by project authors according to these instructions using this template.

Introduction

A total of 42 PhD students in 13 groups did projects of their choosing in Scalable Data Science and Distributed Machine Learning, a mandatory as well as elective course of The WASP Graduate School in 2022-2023. See ScaDaMaLe Course Pathways to appreciate the pre-requisite modules 000_1 through 000_9 for the union of all 13 projects.

The Best Student Group Projects on the basis of peer-review and industrial feed-back are:

  • Group 5 on Scalable Bayesian optimization with distributed Gaussian processes and deep kernel learning (academic-track)
    • Carl Hvarfner, Lund University
    • Leonard Papenmeier, Lund University
    • Manu Upadhyaya, Lund University
  • Group 9 on Predicting the load in wireless networks (industry-track)
    • Sofia Ek, Department of Information Technology, Uppsala University
    • Oscar Stenhammar, Network and System Engineering, KTH and Ericsson

Table of Contents

  1. Graph of Wiki by Vilhelm Agdur, Henrik Ekström, Simon Johansson and Albin Toft.
  2. Visual Question Answering using Transformers by Ehsan Doostmohammadi and Hariprasath Govindarajan.
  3. Scalable Analysis of a Massive Knowledge Graph by Filip Cornell, Yifei Jin, Joel Oskarsson and Tianyi Zho.
  4. Federated Learning for Brain Tumor Segmentation by Jingru Fu, Lidia Kidane and Romuald Esdras Wandji.
  5. Scalable Bayesian optimization with distributed Gaussian processes and deep kernel learning by Carl Hvarfner, Leonard Papenmeier and Manu Upadhyaya.
  6. Experiments with ZerO initialisation by Livia Qian and Rajmund Nagy.
  7. Smart Search in Wikipedia by David Mohlin, Erik Englesson and Fereidoon Zangeneh.
  8. Distributed Ensembles for 3D Human Pose Estimation by Hampus Gummesson Svensson, Xixi Liu, Yaroslava Lochman and Erik Wallin.
  9. Predicting the load in wireless networks by Sofia Ek and Oscar Stenhammar.
  10. Collaborative Filtering in Movie Recommender Systems by Jacob Lindbäck, Rebecka Winqvist, Robert Bereza and Damianos Tranos.
  11. Federated Learning Using Horovod by Amandine Caut, Ali Dadras, Hoomaan Maskan and Seyedsaeed Razavikia.
  12. Distributed Reinforcement Learning by Johan Edstedt, Arvi Jonnarth and Yushan Zhang.
  13. Earth Observation by Daniel Brunnsåker, Alexander H. Gower and Filip Kronström.
  14. Conclusion and BrIntSuSb by Raazesh Sainudiin.
  15. Editors

Invited Talks from Industry

Thanks to the inspiring talks from the following invited speakers from industry:

  • Vivian Ribeiro, Nanxu Su and Tomas Carvalho, trase (Stockholm, Sweden), Transparency for sustainable trade.
  • Reza Zadeh, Matroid and Stanford University (Palo Alto, California, USA), Computer Vision from an academic perspective.
  • Andreas Hellander, Scaleout Systems and Uppsala University (Uppsala, Sweden), Taking Federated Learning to Production - towards privacy-preserving ML at scale.
  • Ali Sarrafi, Christian Von Koch and William Anzen, Combient Mix (Stockholm, Sweden), Slag segmentation with deep neural networks at LKAB.
  • Juozas Vaicenavicius, SENSmetry (Uppsala, Sweden and Vilnius, Lithuania), Autonomous systems safety: what is so difficult?
  • Jim Dowling, Logical Clocks, hopsworks and KTH Royal Institute of Technology (Stockholm, Sweden), Serverless Machine Learning with Hopsworks.