3_objects.md

Working with objects

How to send different shaped data into net?

If we have indexed array it may look like this and be pretty simple:

But what about Objects?

Let's try that:

Input: { red, green, blue } Output: { light, neutral, dark}

Now let's do js out of it:

1. Set the I and O:

const colors = [
  { green: 0.2, blue: 0.4 },
  { green: 0.4, blue: 0.6 },
  { red: 0.2, green: 0.8, blue: 0.8 },
  { green: 1, blue: 1 },
  { red: 0.8, green: 1, blue: 1 },
  { red: 1, green: 1, blue: 1 },
  { red: 1, green: 0.8, blue: 0.8 },
  { red: 1, green: 0.6, blue: 0.6 },
  { red: 1, green: 0.4, blue: 0.4 },
  { red: 1, green: 0.31, blue: 0.31 },
  { red: 0.8 },
  { red: 0.6, green: 0.2, blue: 0.2 }
];

const brightnesses = [
  { dark: 0.8 },
  { neutral: 0.8 },
  { light: 0.7 },
  { light: 0.8 },
  { light: 0.9 },
  { light: 1 },
  { light: 0.8 },
  { neutral: 0.7, light: 0.5 },
  { dark: 0.5, neutral: 0.5 },
  { dark: 0.6, neutral: 0.3 },
  { dark: 0.85 },
  { dark: 0.9 }
];

Don't be so scared, we are just assigning them to each other, so we feeding net a table like this:

red	green	blue	dark	neutral	light
0	0.2	0.4	0.8	0	0
0	0.4	0.6	0	0.8	0
0.2	0.8	0.8	0	0	0.7
0	1	1	0	0	0.8
0.8	1	1	0	0	0.9
1	1	1	0	0	1
1	0.8	0.8	0	0	0.8
1	0.6	0.6	0	0.7	0.5
1	0.4	0.4	0.5	0.5	0
1	0.31	0.31	0.6	0.3	0
0.8	0	0	0.85	0	0
0.6	0.2	0.2	0.9	0	0

2. Now we parse it into trainingData in I/O format:

const trainingData = [];

for (let i = 0; i < colors.length; i++)
  trainingData.push({ input: colors[i], output: brightnesses[i] });

3. and setup the network:

const net = new brain.NeuralNetwork({ hiddenLayers: [3] });

const stats = net.train(trainingData);

console.log(stats);

That's it. Network is learned in 1500 errors.

Now we can ask it, what does it think of some color:

console.log(
  net.run({
    red: 0.9
  })
);

Fun fact: if we want net to give us colors from params, we can just invert I and O:

const invertedTrainingData = [];

for (let i = 0; i < colors.length; i++) {
    invertedTrainingData.push({
        input: brightnesses[i],
        output: colors[i]
    });
}

const invertedNet = new brain.NeuralNetwork({ hiddenLayers: [3] });

const invertedStats = invertedNet.train(invertedTrainingData);

and it fails, but that's not the point. The point is how they work