#evolutionaryalgorithms — Public Fediverse posts

DeepMind's AlphaEvolve strikes again:

After recovering 0.7% of global GCP compute through better scheduling algos, AlphaEvolve helped FM Logistics achieve "10.4% improvement in routing efficiency over the previous best solution", leading to "15,000+ fewer kilometers of warehouse travel per year at full operational scale".

It's really nice to see evolutionary algorithms make a comeback. AlphaEvolve uses an LLM (Gemini) as as a mutation function, and using similar mutate-evaluate-select loops typical of traditional evolutionary algorithms.

https://cloud.google.com/blog/products/ai-machine-learning/how-fm-logistic-tackled-the-traveling-salesman-problem-at-warehouse-scale-with-alphaevolve

#AlphaEvolve #DeepMind #EvolutionaryAlgorithms

I'm really enjoying my latest research project. This one's exploring how different spatial environments can lead to different evolutionary dynamics. Here we see an environment where it's harder to survive in the middle than on the edges (that is, it requires higher scores from a fitness function). We can see the population evolve increasing fitness as it spreads into the interior space.

#EvolutionaryComputation #EvolutionaryAlgorithms #science #evolution

Just checked #EvolutionaryAlgorithms and #EvolutionaryComputation tags. And I found posts from just two: myself and @moshesipper

It's time to feel scientifically lonely :)

But really, why some people like me are attracted to less or more unpopulated regions of science and engineering?

#science

New paper out ✒️😊

We present a novel approach to performing fitness approximation in #geneticalgorithms (#GAs) using #machinelearning (#ML) models, focusing on dynamic adaptation to the evolutionary state.

https://www.mdpi.com/2078-2489/15/12/744

With talented grad students Itai Tzruia and Tomer Halperin, and my colleague Dr. Achiya Elyasaf.

#evolutionaryalgorithms

#evolutionarycomputation

One idea I'm thinking about for #GeneticProgramming

In GP, we usually need to evolve multiple subpopulations named "demes" rather than one big population as the converge on smaller population happens faster. Then we have small, infrequent migrations between subpopulations, like 5% of population of each subpopulation/deme.

#GP #GeneticProgramming #EvolutionaryAlgorithms #EA #GeneticAlgorithm #EvolutionaryMachineLearning #AI #ML #EML #research #researching #Question #Idea

1/2