What to do: 1. Choose your input 2. Choose your goal (expected) Value that it has to achieve (between 0 and 1) -> Program tries to set OVal to this nuber (list in the middle on th right) -> please choose a number bigger then 0.1 or it will sometimes break, but if you do, set the learning value very smal (like 0.0001 or 0.000001) and make a few 1000 or 10000000 Generations. For faster loading, use Turbomode (Shift+Green Flag). 3. Set the Learning Rate (best between 0.001 and 0.01) -> Smaller = slower, but more precise 4. Choose how many Generations to simulate (about 300-1000) -> not to many or the number will overshoot Explination: After you set everything up, the program tries to set the Output Number (in the OVal list) as close as possible to your expected number. It does so by changing its own settings (weights and bias) to calculate the same number as you have choosen. How close the program is to your number is indicated by the Loss Variable. The closer it is to 0, the close are the numbers. In the end it means that the network learns, that the input number you have choosen is equal to your expected number. Notes: This is my 1. attempt on making a Neural Network, the base of an AI. Because this is the 1. version, it only has 1 input, 1 hidden and 1 output neuron, but it can still learn to get closer to your choosen (expected) number (only works with numbers between 0 and 1 right now). For the Learning Rate, I recomend a number between 0.001 and 0.01.If the Learning Rate is to small, it wont be able to come very close to your choosen number, but if it is to big, it will completely break and show something like -infinity after a few generations. There are still a few bugs like the Loss becoming bigger again after a few hundred generations, but i try to fix them soon. In the future I want to make an image identifier or maybe even generator out of this base program.
Changelog: -Updated/fixed all Formulas -> less random jumps -Added 1 hidden neuron -added output visualizer -Updatet interface -halfautomiced weightconnection drawing V1: https://scratch.mit.edu/projects/1322705550/ Thanks to 3Blue1Brown for the Youtube explination series: https://www.youtube.com/watch?v=aircAruvnKk&list=PLZHQObOWTQDNU6R1_67000Dx_ZCJB-3pi Note: It is still very buggy, but i will fix it soon
