Added examples

.gitignore

@@ -0,0 +1 @@
+node_modules/

examples/XOR.js

@@ -0,0 +1,30 @@
+const NeuralNetwork = require("neural-network-node");
+
+let nn = new NeuralNetwork(2, 4, 1);
+
+let training_data = [
+  {
+    inputs: [0, 0],
+    outputs: [0],
+  },
+  {
+    inputs: [0, 1],
+    outputs: [1],
+  },
+  {
+    inputs: [1, 0],
+    outputs: [1],
+  },
+  {
+    inputs: [1, 1],
+    outputs: [0],
+  },
+];
+
+for (let i = 0; i < 1000; i++) {
+  let data = random(training_data);
+  nn.train(data.inputs, data.outputs);
+}
+
+const output = nn.predict([0, 1]); //The more you train the model, the closer this gets to one
+console.log(output);

examples/sinwave.js

@@ -0,0 +1,23 @@
+const NeuralNetwork = require("neural-network-node");
+
+let nn = new NeuralNetwork(1, 2, 1);
+
+function generatedata() {
+  let input = Math.random() * 2 * Math.PI;
+  let output = Math.sin(input);
+
+  return {
+    input: [input],
+    output: [output],
+  };
+}
+
+for (let i = 0; i < 10000; i++) {
+  var data = generatedata();
+  nn.train(data.input, data.output);
+}
+
+console.log("Finished training");
+
+const output = nn.predict(Math.PI); //The more you train the model, the closer this gets to zero
+console.log(output);

package.json

@@ -1,6 +1,6 @@
 {
   "name": "neural-network-node",
-  "version": "1.0.4",
+  "version": "1.1.0",
   "description": "A simple 2 layer neural network coded from scratch with no external modules. Lets you train your own neural network in a simple and easy to learn way.",
   "main": "src/index.js",
   "scripts": {
@@ -13,8 +13,18 @@
     "node",
     "javascript",
     "machine",
-    "learning"
+    "learning",
+    "supervised",
+    "data"
   ],
   "author": "SatvikVejendla",
-  "license": "ISC"
+  "license": "ISC",
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/SatvikVejendla/Neural-Network-Node.git"
+  },
+  "bugs": {
+    "url": "https://github.com/SatvikVejendla/Neural-Network-Node/issues"
+  },
+  "homepage": "https://github.com/SatvikVejendla/Neural-Network-Node#readme"
 }

utils/NeuralNetwork.js

@@ -19,8 +19,12 @@ class NeuralNetwork {
     this.learning_rate = 0.1;
   }
 
+  getLearningRate() {
+    return this.learning_rate;
+  }
+
   setLearningRate(x = 0.1) {
-    this.learning_rate = learning_rate;
+    this.learning_rate = x;
   }
 
   setWeights(x) {
@@ -48,7 +52,13 @@ class NeuralNetwork {
     }
   }
 
-  feedforward(input_array) {
+  predict(input_array) {
+    if (input_array.length != this.input_nodes.length) {
+      const err =
+        "ERROR! The predict function input array does not match the length of the neural network's input nodes.";
+      console.log(err);
+      return err;
+    }
     let inputs = Matrix.fromArray(input_array);
 
     //input to hidden
@@ -68,46 +78,59 @@ class NeuralNetwork {
   }
 
   train(input_array, target_array) {
-    let inputs = Matrix.fromArray(input_array);
-
-    //input to hidden
-    let hidden = Matrix.multiply(this.weights_ih, inputs);
-    hidden.add(this.bias_h);
-
-    //activation
-    hidden.map(Sigmoid.sigmoid);
-
-    //hidden to output
-    let outputs = Matrix.multiply(this.weights_ho, hidden);
-    outputs.add(this.bias_o);
-    outputs.map(Sigmoid.sigmoid);
-
-    let targets = Matrix.fromArray(target_array);
-    //error calculation
-    let output_errors = Matrix.subtract(targets, outputs);
-    let gradients = Matrix.map(outputs, Sigmoid.dsigmoid);
-    gradients.multiply(output_errors);
-    gradients.multiply(this.learning_rate);
-
-    let hidden_T = Matrix.transpose(hidden);
-    let weight_ho_deltas = Matrix.multiply(gradients, hidden_T);
-
-    //Weights and Biases adjustments
-    this.weights_ho.add(weight_ho_deltas);
-    this.bias_o.add(gradients);
-
-    //hidden layer errors
-    let who_t = Matrix.transpose(this.weights_ho);
-    let hidden_errors = Matrix.multiply(who_t, output_errors);
-    let hidden_gradient = Matrix.map(hidden, Sigmoid.dsigmoid);
-    hidden_gradient.multiply(hidden_errors);
-    hidden_gradient.multiply(this.learning_rate);
-
-    let inputs_T = Matrix.transpose(inputs);
-    let weight_ih_deltas = Matrix.multiply(hidden_gradient, inputs_T);
-
-    this.weights_ih.add(weight_ih_deltas);
-    this.bias_h.add(hidden_gradient);
+    if (input_array.length != this.input_nodes.length) {
+      console.log(input_array.length + ": " + this.input_array.length);
+      const err =
+        "ERROR! The train function input array does not match the length of the neural network's input nodes.";
+      console.log(err);
+      return err;
+    } else if (target_array.length != this.output_nodes.length) {
+      const err =
+        "ERROR! The train function target output array does not match the length of the neural network's output nodes.";
+      console.log(err);
+      return err;
+    } else {
+      let inputs = Matrix.fromArray(input_array);
+
+      //input to hidden
+      let hidden = Matrix.multiply(this.weights_ih, inputs);
+      hidden.add(this.bias_h);
+
+      //activation
+      hidden.map(Sigmoid.sigmoid);
+
+      //hidden to output
+      let outputs = Matrix.multiply(this.weights_ho, hidden);
+      outputs.add(this.bias_o);
+      outputs.map(Sigmoid.sigmoid);
+
+      let targets = Matrix.fromArray(target_array);
+      //error calculation
+      let output_errors = Matrix.subtract(targets, outputs);
+      let gradients = Matrix.map(outputs, Sigmoid.dsigmoid);
+      gradients.multiply(output_errors);
+      gradients.multiply(this.learning_rate);
+
+      let hidden_T = Matrix.transpose(hidden);
+      let weight_ho_deltas = Matrix.multiply(gradients, hidden_T);
+
+      //Weights and Biases adjustments
+      this.weights_ho.add(weight_ho_deltas);
+      this.bias_o.add(gradients);
+
+      //hidden layer errors
+      let who_t = Matrix.transpose(this.weights_ho);
+      let hidden_errors = Matrix.multiply(who_t, output_errors);
+      let hidden_gradient = Matrix.map(hidden, Sigmoid.dsigmoid);
+      hidden_gradient.multiply(hidden_errors);
+      hidden_gradient.multiply(this.learning_rate);
+
+      let inputs_T = Matrix.transpose(inputs);
+      let weight_ih_deltas = Matrix.multiply(hidden_gradient, inputs_T);
+
+      this.weights_ih.add(weight_ih_deltas);
+      this.bias_h.add(hidden_gradient);
+    }
   }
 }