fksainetwork
A neural network capable of learning and calculating output. This project is just made for fun :)
Please go to Inner Workings to read more.
Usage
[]
= "0.2.0"
Example (Feed Forward)
let mut network = new;
//or: let network = load_network("path/to/network-file");
//calculating
let input = vec!;
let output = network.calculate; //calculate
println!;
//learning
//batch size of 2
network.learn;
//NOTE: if you call 'learn', u do not need to call 'calculate' beforehand
//save
save_network;
Example (Convolutional)
let network = new;
//or: let network = load_cnn_network("path/to/network-file");
//pretend these samples are actual images of something
let sample0 = new;
let sample1 = new;
let sample2 = new;
//calculate
let output = network.calculate;
println!;
//learn
network.learn;
//print outputs
cnn_network_bmp;
//save network
save_cnn_network;
Update Patches :)
0.1.3: Added Tanh Activation Function. 0.1.3: Improved Docs. 0.2.0: The save_network and load_network functions now use str instead of String for the path. 0.2.0: Improved learning algorithm. 0.2.0: Added the Convolutional Neural Network. 0.2.0: Added more Loss functions and Initialization functions. 0.2.0: Changed learn_bpg_mse to learn as the loss function is now a function parameter. 0.2.0: Changes to the feed forward network "new()" function parameter.