💾 Archived View for envs.net › ~arydev › projectAH › posts › 1.gmi captured on 2021-12-04 at 18:04:22. Gemini links have been rewritten to link to archived content

View Raw

More Information

-=-=-=-=-=-=-

Back

Home

Title: The Brain!

Date: September 21 2021

Now the thing with this project is to advance

AI tehnology, so i won't use any existing algorithm,

also because it won't work well with my project.

I am going to use a custom algorithm inspired from

Neutral Networks and Reinforcement Learning.

It is split in 3 parts, input, logic and output layer

The input and output layers are the same as neutral networks,

while the logic layer is similar to the deep layers,

the logic layer is made of nodes that sit at random

positions (kind of) and they can connect to any input,

output or any other node,the number of nodes is unlimited,

it is set at creation but in the evolution it can change.

Instead of biases and stuff, my algorithm uses logic gates,

this makes it easier for a human to understand it.

Now here is the cherry on the cake, memory.

The most useful thing about this is that

it uses a tree based memory, the algorithm

is the best when it is made of multiple

instances that talk to each other.The algorithm

sends a number to the memory module that based

on the binaty representation jumps left or right,

using the response it generate a output, and

the next instance uses the memory output and

the last instance output to generate another one.

This keeps until the instances end, and a final

response is outputed.

Hopefully the next week i am gona post about my first

prototype.Bye, for now.

(yes i am bad at writing blogs)