Evolutionary Algorithms II

The seminar consists of presentations on different selected topics prepared by the students. Ideally, each lesson will contain four presentations, thus, each presentation should be at most 20 minutes including questions and discussion.

You can come up with your topic, or you can choose one from the list bellow. There are two options regarding the content of the presentation: you can either choose an interesting application of evolutionary algorithms, implement it and present the results, or you can “only” study some interesting topic and tell us about it. If the selected topic is also included in the lecture, the presentation should contain something new (interesting application, some new algorithm, etc.). Repeating what Roman said during the lecture does not make sense.

Credit requirement

Choose the topic during the first week of the term and also choose the date for your presentation. Put them into the shared table (I sent you a link, or you can find it as a comment to the schedule in sis).
Upload your slides (or other notes, if you do not use slides) to the shared directory I sent you in an e-mail (the link is also in the document with the shared table).
Attend the seminar whenever possible.
You can also work in pairs, but in such a case, you should also have an implementation and demonstration as a part of the presentation

Calendar of the seminar

The selected topics and free slots for presentations are in a shared table. I sent you a link on February 16, 2025. It is also in a note in schedule in sis. If there is no presentation planned for a given day, the seminar is canceled.

List of topics

This list is incomplete, you can choose any other topic, which may be interesting for others and is at least a little connected to natural or artificial evolution, artificial life, or related fields. The links should provide some basic information about the topic, but you should use other sources in the presentation.

Another good source for topic can be papers published recently in evolutionary computation conferences (GECCO, IEEE CEC) or in machine learning or AI conferences (ICML, ICLR, IJCAI). In these conferences, you should focus on applications of evolutionary algorithm in areas such as neural architecture search or automated machine learning.

Applications of evolutionary algorithms

Genetic Programming — relatively wide topic, also discussed in the lecture, but there are definitely topics that can be used for presentation (e.g. semantic GP, linear GP, cartesian GP, etc.)
Graph Drawing — the task is to draw a graph in such a way that no edges cross (or minimize the number of crossing edges)
Co-evolution — both natural and artificial
Neuro-evolution — again a wide topic, partially discussed in the lectures, but e.g. EANT can be used for presentation
Scheduling — wide topic, which is not discussed in the lecture, you can also compare evolutionary and classical techniques
Evolutionary Art — music and pictures created by evolution (or TORCS race courses)

Theory of evolutionary computation

Swarm Intelligence — an optimization technique inspired by the behavior of swarms
Ant Colony Optimization — optimization algorithm inspired by ants
Harmony search — inspired by the improvisation in music
Artificial Immune Systems — inspired by the Immune System
Estimation of Distribution Algorithms — algorithms, which generate new individuals from an estimated probability distribution
Evolutionary Theories — including old ones, Lamarckism, Baldwinism, Darwinism, but other types of evolution can also be interesting — socio=cultural, technological.
Artificial Life — both software- and hardware-based

Books about evolution

Richard Dawkins — Selfish Gene — Dawkins’s theory of selfish gene
Jaroslav Flégr — Frozen evolution — theory of frozen evolution, how bottlenecks lead to evolution
Matt Ridley — Red Queen — about sexuality and sexual selection