Let me write my first thread here about #binaryTrees and how #algorithms can differ from case to case.

So I wrote, in #python, 2 different implementations for an Integer Binary Tree:
- Recursive
- Iterative

You can check here the code: https://gist.github.com/ttiagojm/526b3f3634dbc751f693c81ae1e20996

And as you can see there is no gain, sometimes even recursion is faster than iterative.

Then I was intrigued by this result and I tried to run an identical test on a project.
1/2

A simple Integer Binary Tree implementation using both Iterative and Recursive implementation

A simple Integer Binary Tree implementation using both Iterative and Recursive implementation - bin_tree.py

Gist

Well, that project implements a binary tree to represent a Convolutional Neural Network.
It creates a random network, symbolically, and evaluates transforming it in a real #Keras model.

Testing on this gave a different result as you can see in the image above. Iterative mode is better and you don't get recursion errors because of recursion limits.

Recursion is more elegant and I like to think recursively, but normally iterative wins :)

2/2

@tiago_j_m Hey do you know about #combinatorial #species? I've been researching these marvelous gizmos for a few years, and tying their development into larger theoretical mosaics. They are closely related to advances in #evolutionary/#genetic and #differentiable #programming.

@yogoy Nope, I started to learn about Evolutionary Algorithms and Genetic Programming a few months ago because I need them for a project. So I don't know many stuff, yet :)

But that seems to be interesting, did you write a paper about that topic?

@tiago_j_m I haven't published anything, and with the recent discourse in the Securities and Exchange Commission I am glad I didn't. Whitepapers themselves may soon be taken as suggestions of criminal conduct.

But, my work touches on a lot more than 500 glyphs can encapsulate! I'd love to learn more about your current interests so I can explain more precisely. I'm into compiler design, distributed and extended intelligence, asynchronous computing etc

@yogoy For now I'm only interested to use evolutionary algorithms to search for optimal Neural Nets architectures/hyperparameters
@tiago_j_m Fantastic! That's pretty much the only thing I care about these days, too. Category theory has offered some delightful ways of conceptualizing operative networks. Graphs are the way.
@yogoy Can you give me some examples of Category Theory applied to NN architecture search?
@yogoy Graphs are almost always the way 😂