From d9ce4b0d70476bc4a7b39161e25fd2d036a5d0e9 Mon Sep 17 00:00:00 2001
From: Ariel Sanguinetti <arielsanguinetti@gmail.com>
Date: Sun, 11 Feb 2018 17:25:20 -0300
Subject: [PATCH] Add simple NN test

---
 lib/nn.js      | 24 +++++++++++++++++-------
 lib/nn.test.js | 37 +++++++++++++++++++++++++++++++++++++
 2 files changed, 54 insertions(+), 7 deletions(-)
 create mode 100644 lib/nn.test.js

diff --git a/lib/nn.js b/lib/nn.js
index 5cce746..0383e73 100644
--- a/lib/nn.js
+++ b/lib/nn.js
@@ -1,7 +1,9 @@
-// Other techniques for learning
+if (typeof(require) !== 'undefined') {
+  Matrix = require('./matrix');
+}
 
-class ActivationFunction{
-  constructor(func, dfunc){
+class ActivationFunction {
+  constructor(func, dfunc) {
     this.func = func;
     this.dfunc = dfunc;
   }
@@ -9,12 +11,12 @@ class ActivationFunction{
 
 let sigmoid = new ActivationFunction(
   x => 1 / (1 + Math.exp(-x)),
-  y => y * (1- y)
+  y => y * (1 - y)
 );
 
 let tanh = new ActivationFunction(
   x => Math.tanh(x),
-  y => 1-(y*y)
+  y => 1 - (y * y)
 );
 
 
@@ -128,8 +130,7 @@ class NeuralNetwork {
   }
 
   static deserialize(data) {
-    if(typeof data == 'string')
-    {
+    if (typeof data == 'string') {
       data = JSON.parse(data);
     }
     let nn = new NeuralNetwork(data.input_nodes, data.hidden_nodes, data.output_nodes);
@@ -142,3 +143,12 @@ class NeuralNetwork {
   }
 
 }
+
+if (typeof module !== 'undefined') {
+  module.exports = {
+    NeuralNetwork: NeuralNetwork,
+    ActivationFunction: ActivationFunction,
+    sigmoid: sigmoid,
+    tanh: tanh,
+  };
+}
diff --git a/lib/nn.test.js b/lib/nn.test.js
new file mode 100644
index 0000000..ea7aa1b
--- /dev/null
+++ b/lib/nn.test.js
@@ -0,0 +1,37 @@
+//const Matrix = require('./matrix');
+const {
+  NeuralNetwork,
+} = require('./nn');
+
+test('XOR Feed forward 100 times (NN [2,4,1], lr = 0.1)', () => {
+  const defaultNN = '{"input_nodes":2,"hidden_nodes":4,"output_nodes":1,"weights_ih":{"rows":4,"cols":2,"data":[[-0.67573921323278,-0.6631309040105413],[0.12043373460672324,0.9164706630341213],[0.8065050836856171,0.6200864007638192],[0.17026959573001887,0.8558180147448349]]},"weights_ho":{"rows":1,"cols":4,"data":[[0.25208907823805715,-0.15513622033337615,-0.7593404409594813,-0.2200931714640979]]},"bias_h":{"rows":4,"cols":1,"data":[[-0.35555433502368805],[0.5915246859082983],[0.7931155122071059],[-0.5196051276393758]]},"bias_o":{"rows":1,"cols":1,"data":[[0.16280502479816095]]},"learning_rate":0.1,"activation_function":{}}';
+
+  const training_data = [{
+      inputs: [0, 0],
+      outputs: [0]
+    },
+    {
+      inputs: [0, 1],
+      outputs: [1]
+    },
+    {
+      inputs: [1, 0],
+      outputs: [1]
+    },
+    {
+      inputs: [1, 1],
+      outputs: [0]
+    }
+  ];
+
+  const targetNN = '{"input_nodes":2,"hidden_nodes":4,"output_nodes":1,"weights_ih":{"rows":4,"cols":2,"data":[[-0.569756962028276,-0.5551249563852504],[0.12350867790872079,0.9117849077336435],[0.6448544388169963,0.37673042356680453],[0.15511871409391612,0.8289329770778586]]},"weights_ho":{"rows":1,"cols":4,"data":[[0.25671912531734464,0.02490177829691785,-0.537758005494704,-0.10072084075558174]]},"bias_h":{"rows":4,"cols":1,"data":[[-0.28493369112696326],[0.6027906492507047],[0.7374757365358221],[-0.5460077788408344]]},"bias_o":{"rows":1,"cols":1,"data":[[0.37380624852136013]]},"learning_rate":0.1,"activation_function":{}}';
+
+  let nn = NeuralNetwork.deserialize(defaultNN);
+
+  for (let i = 0; i < (training_data.length * 100); i++) {
+    let index = i % training_data.length;
+    let data = training_data[index];
+    nn.train(data.inputs, data.outputs);
+  }
+  expect(nn.serialize()).toEqual(targetNN);
+});