eheheh the article say that another article method spent many time to find an architecture for a small #NeuralNetwork, because they used only mutation on they #evolutionaryalgorithm

BUT the article had 186 TFLOPs (9 days of searching) while the other had only 18 TFLOPs (14 days of searching)

Have you ever seen an #EvolutionaryAlgorithm evolve to "extinction"? By that I mean that fitness is gradually improving over multiple generations and then suddenly falls off a cliff to a lower bound, and never recovers. A student saw it in some code we were hacking recently. The cause was obvious, in retrospect.

Hey all, don't know what's all the fuss, I'm not even that aware of what's exactly going on twitter, but if even 10% of what I've been reading is true it's probably a smart move to split my eggs across baskets 🤔

For the ones who don't know, my name is Léo and I've been developing The Bibites, an #ArtificialLife project, for the past ~7 years.

My goal is to keep pushing the boundaries of Artificial Life Simulations forward and to never stop never stopping

#introduction #evolutionaryalgorithm

A #thread of recent papers coauthored by Eörs Szathmáry on learning and evolution, covering Darwinian principles in the immune system, brain, Bayesian learning theory.

Mentioned in today's #KLIAustria #KLIColloquium talk.

#evolution #ecoevo #EvolutionaryBiology #EvolutionaryAlgorithm #Darwin #Darwinian #Learning #LearningTheory #Bayesian #neuroscience #immunology #PaperThread #

Today's talk at the #KLI is about the research behind Eörs Szathmáry's collaboration with Richard A. Watson.

How Can Evolution Learn?
https://doi.org/10.1016/j.tree.2015.11.009

#evolution #EvolutionaryBiology #EvolutionaryAlgorithm #Learning #ecoevo

@TrendsEcolEvo

Szathmáry is a senior fellow & board member of #KLIAustria.

https://www.kli.ac.at/en/events/event_calendar/view/646