Visit the tool: galinator9000.github.io/artificialneural.network
This is an interactive web page that runs a toy neural network right on your browser!
It visualizes the learning processes of neural networks; the building block of the AI revolution we are witnessing today.
This tool utilizes libraries like p5.js for being able to visualize the neural network and tensorflow.js in order to run the underlying mechanisms of them.
To get started;
-
Choose a sample dataset from the dataset section and click on 'Compile Dataset'
-
If you'd like to, you can fully configure your neural network from the 'Network Structure' part on right-bottom. However, the tool will set a default network configuration that depends on the chosen dataset's problem to get you started.
-
Click on "Compile network" for initializing the network with random parameters.
-
You can then start the learning process, either with the single sample it has chosen from the dataset or make it learn the whole dataset.
You can click on any of the neurons to see how strong their connections are and how they relate with each other.
Play as you like!