(The life and times of me keeping up with the things that should be constant in a sea of variables...)
my next coding project...
is not the really a project that I've really started, kinda pumped the breaks on it. IT'S JOB HUNTING SZN and stuff isn't progressing the way it's supposed to. I'm nearing the end of my staying period and jobs aren't coming by man. Life is hard.
So currently I'm just going through some fundamental/machine-learning algorithms. Currently on Mixture Models and Hash Maps. Nothing fancy I know but times are tough man, gotta survive.
On the other hand, I do have a Neural Net Project in mind. So i saw a video where an AI was playing flappy bird and crushing it. That gave me an idea to create an AI that could handle these simple decision based games via some neural net.
So first up I'm contemplating the Chrome dinosaur game. The needs
are minimal cause it's got only one button probably 4-5 kinds of
jumps, that's 4 kinds of input to be timed, and well the rest is
followed by how the learning weights are designed.
I'd have to use probably Selenium to automate the browser
commands, or maybe Google's Tensorflow.js for adding machine
learning capabilities to a back-end and browser. However I might have
to have a separate simulation of the game rather than the traditional
"bruh you're internet ain't workin." to make it run.
Hmm now that I think of it, even the crash data has to be collected
push these in the training set to be input back into the training model.
Until all aspects, i.e, possible jumps from all possible cacti distance,
aren't covered there is always going to be a crash.
Same theory applies for the crouching parts.
As for the activation function, We can probably go for a Sigmoid function. Pretty easy. or not. somebody help my anxiety.
....Gawwdddamn son now im too deep in this. Gotta get back to
a job. :(
i will be pushing this to my Github for sure. Gotta make it open source so I can learn a thing or two and watch the world upgrade my code like Thanos just got his 6th championship ring. Right now I'll be adding some multivariate logistic regression code and probably bagging/boosting models.
ARCHIVE {AVAILABLE ON my GitHub}
(2020)
Snake using vanilla JS
LeetCode solutions
.
.
(2019)
Instagram App for Stalking-ish User Functions (check my Github)
Instagram App for building Bots to well....increase your like and followers you desperate clown.
LeetCode solutions
