Differentiable Genetic Programming (2016)

arxiv.org

31 points·henning··7 comments

I'm still facepalming GP because the associative and communative properties of algebra are omitted. Add in a nonlinear solver and you can reduce th problem size considerably.

http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.394....

So do you mean that GP could sped-up by taking the evolutionary process it's simulating and using a solver to find the process' fixed-points?

GP in general needs something like Tensorflow and a standard set of benchmarks

Well it needs something, it seems.

The article looks like an interesting approach but it gives every impression of being a "cul-da-sac", an effort that wasn't followed up on. Google scholar shows this article cited by four other articles - and those article not cited by anything more.

Some way of figuring out what articles matter in GP would be very useful.

gpbenchmarks.org ?

That's ok, there was a paper from Gecco2012, I think, GP needs better benchmarks.

Differential Equations and PDEs would be a good option too.

The fact that there is no standard set of benchmarks was recognised some years ago:

https://cs.gmu.edu/~sean/papers/gecco12benchmarks3.pdf

http://www.cs.put.poznan.pl/wjaskowski/archives/wp-publicati...

While I think that the situation is improving, we are still far from a common set of benchmarks used by the entire GP community.

Reply on news.ycombinator.com