Machine Learning Experiments
The first taken from a tutorial
It works, but simply memorizes the map, so doesn't work on any map.
My first attempt to modify QMaze to work with multiple maps. A failure.
My first succesful attempt to work with multiple maps. Still not sure yet if it is just memorizing multiple maps.
An attempt to learn to use LSTM - long-term-short-term memory
Just learning how to use ML to create a book recommendation system
Multi inputs then try to predict the output
Testing custom fit
Experimenting with self categorizing so a system can recommennd similar movies, products or recipies for example
First step towards the schedule generator. A reinforcement learning process where we don't knnow the outputs, just a reward.
Continnuing on from cart-pole, another reinforcement exercise
Given an unknnown function, find the opimum from it
Exploring RL to generate a timetable
The schedule generator sucks at exploration, so this project is setup to help try different explorationn methods
FUTURE IDEAS: weather rain forecast - how lonng bbefore it rains
https://github.com/e-lab/VideoSensors - predict gyroscope data from video footage
noise cancelling with 2 microphones, to determine what is outside sound to block and what is inside sound to ignore
generate website from image (midjourney)
ai to tweak image ready for text recognition for SS
TODO: I realised that the possible problem with my QMaze was that the inputs were absolute, not relative. I need to give the AI the pixels around where they are located in concentric circles. This might also work to give them a history of past steps.