Noon seminar
The informal research seminar of the ALGA group. Talks last roughly 25 minutes, with five extra minutes allocated for discussion. Many presentations are focused on recent conference presentations, or practice talks for upcoming conferences. New members are often asked to give an overview of their field of research. Talks given by invited speakers may take up to 45–60 minutes including questions.
To be kept uptodate about noon seminar presentations, please subscribe to the algoseminarl mailing list.
COVID19 note
We currently conduct talks via Zoom. To avoid abuse, we do not distribute the join links publicly. To make the links appear in the list below, make sure you are signed in. Alternatively, you can find the join link in the announcement email.
This is the schedule for 2021. Schedules are available between 2005 and 2021.
 Presentation
 Master's defense
 Midterm
 Invited talk
 External talk
 Practice talk
 PhD defense
 Cancelled
Upcoming talks
Send an email to Alex to schedule your talk.
No talks are planned at the moment.
Past talks

Apr 12, 10:00—10:40
(Zoom)
Remco Surtel: Leader Election in the Amoebot Model.
－ Abstract
Programmable matter is a concept in the field of distributed computing, it refers to a collection
of synthetic particles which can be programmed to change the physical properties of the material
that they form. Leader election is a common and important problem in distributed computing,
in which a single node must be chosen to lead the execution of a larger algorithm. In the context
of programmable matter, leader election is a significantly complex problem, both due to the large
numbers of particles in a system, and the storage and computational limitations of the individual
particles. In this thesis, several leader election algorithms for particle systems of programmable
matter are evaluated both experimentally and analytically. We consider the Amoebot model for
programmable matter, which describes how particles can move, compute, and interact with one
another. Four different leader election algorithms are implemented in AmoebotSim, a simulator
for the Amoebot model. Using experiments we determine which of these algorithms is fastest,
and what factors influence the running time of each algorithm. Furthermore, we investigate
the theoretical robustness and probability of failure of the algorithms where applicable. This
information should be helpful for the selection of a leader election algorithm on given types of
input particle systems, and may also help in the development of new algorithms with different
performance characteristics or assumptions.

Apr 06, 11:00—11:30
(Zoom)
Pantea Haghighatkhah: EuroCG Practice Talks: Obstructing Classification via Projection.
－ Abstract
EuroCG has 12minute talks with 3minute breaks for questions and switching. We schedule 15 minutes extra for feedback and inevitable technical issues.

Mar 31, 15:00—16:30
(Zoom)
Aleksandr Popov, Bram Custers, and Nathan van Beusekom: EuroCG Practice Talks.
－ Abstract
EuroCG has 12minute talks with 3minute breaks for questions and switching. We schedule 15 minutes extra per person for feedback and inevitable technical issues.
1. Aleksandr Popov – Uncertain Curve Simplification
2. Bram Custers – Route Reconstruction from Traffic Flow via Representative Trajectories
3. Nathan van Beusekom – Crossing Numbers of BeyondPlanar Graphs Revisited

Mar 30, 10:00—10:30
(Zoom)
Tom Peters: EuroCG Practice Talks: Coordinating Programmable Matter via Shortest Path Trees.
－ Abstract
EuroCG has 12minute talks with 3minute breaks for questions and switching. We schedule 15 minutes extra for feedback and inevitable technical issues.
Upd: Pantea’s talk has been rescheduled to 6th April.

Mar 23, 13:00—13:40
(Zoom)
Carina Cornet: Map Matching with a Continuous Hidden Markov Model.
－ Abstract
The aim of this thesis is to define to what extent the exact route driven can be determined from a GPS trajectory using a continuous hidden Markov model (CHMM). We present a short overview of the various map matching methods and a more extensive overview of the hidden Markov model (HMM). Current HMMs are discrete as they calculate with the closest position which is not necessarily the most optimal point. Thus, the CHMM defines the measurement and transition probability in a continuous way, by defining a probability distribution over all points. Using data from HERE Technologies we test the performance of the CHMM for various noise levels and sampling intervals. The CHMM outperforms some models for short sampling intervals and has a similar overall performance to others. We observe two main limitations. First, the CHMM fails to take into account the correct route when finding the shortest path between candidates if the sampling interval is too long. Second, if the distance traversed on the road is considerably larger than the distance between consecutive measurements, the transition probability gets too small and the model is unable to find the correct route. The latter limitation may be overcome by implementing a different projection technique for the transition probability. A different projection technique, presented in this thesis, uses time instead of distance, which requires a good estimation of the average velocity, as it is very sensitive to traffic conditions and the accuracy of the recorded velocity of the vehicle. The results show that, besides the limitations, the CHMM is a simple but effective continuous map matching method.