model and training with dummy data

Author: shiffman

index.js

@@ -1,3 +1,58 @@
 console.log("Hello Autoencoder 🚂");
 
 import * as tf from '@tensorflow/tfjs-node'
+
+const autoencoder = tf.sequential();
+
+const encoder = tf.layers.dense({
+  units: 32,
+  inputShape: [784],
+  activation: 'relu'
+});
+const decoder = tf.layers.dense({
+  units: 784,
+  activation: 'sigmoid'
+  // inputShape: [32]
+});
+
+autoencoder.add(encoder);
+autoencoder.add(decoder);
+
+autoencoder.compile({
+  optimizer: 'adam',
+  loss: 'binaryCrossentropy',
+  metrics: ['accuracy'],
+});
+
+
+function generateImage() {
+  const img = [];
+  for (let i = 0; i < 784; i++) {
+    img[i] = Math.random();
+  }
+  return img;
+}
+
+const x_inputs = [];
+for (let i = 0; i < 1000; i++) {
+  x_inputs[i] = generateImage();
+}
+
+const x_train =  tf.tensor2d(x_inputs);
+x_train.print();
+
+trainModel();
+
+async function trainModel() {
+  await autoencoder.fit(x_train, x_train, {
+    epochs: 50,
+    batch_size: 256,
+    shuffle: true,
+    verbose: true
+  });
+}
+
+
+
+
+