neuralnet/network_test.go

package nn

import (
	"gonum.org/v1/gonum/floats"
	"testing"
)

func TestEmptyNetworkErrors(t *testing.T) {

	F255 := make([]uint, 255)
	for i, _ := range F255 {
		F255[i] = 1
	}
	F256 := make([]uint, 256)
	for i, _ := range F256 {
		F256[i] = 1
	}

	tests := []struct {
		Args []uint
		Err  error
	}{
		{[]uint{}, ErrMissingLayers},
		{[]uint{1}, ErrMissingLayers},
		{[]uint{1, 2}, nil},
		{F255, nil},
		{F256, ErrTooMuchLayers},
		{[]uint{1, 1, 0}, ErrEmptyLayer},
	}

	for _, test := range tests {

		_, err := Empty(test.Args...)

		// unexpected error
		if test.Err == nil && err != nil {
			t.Errorf("Unexpected error <%s>", err)
			continue
		}

		// expected error
		if test.Err != nil && err != test.Err {
			t.Errorf("Expected error <%s> but got <%s>", test.Err, err)
			continue
		}

	}

}

func TestEmptyNetworkSizes(t *testing.T) {

	tests := [][]uint{
		{1, 2, 3},
		{4, 6, 9, 10},
		{1, 1, 1},
	}

	for _, test := range tests {

		net, err := Empty(test...)

		// unexpected error
		if err != nil {
			t.Errorf("Unexpected error <%s>", err)
			continue
		}

		// 1. Check neuron layer count
		if len(net.Neurons) != len(test) {
			t.Errorf("Expected %d layers of neurons, got %d", len(test), len(net.Neurons))
			continue
		}

		// 2. Check bias layer count (layers-1)
		if len(net.Biases) != len(test)-1 {
			t.Errorf("Expected %d layers of biases, got %d", len(test)-1, len(net.Biases))
			continue
		}

		// 3. Check weights layer count (layers-1)
		if len(net.Weights) != len(test)-1 {
			t.Errorf("Expected %d layers of weights, got %d", len(test)-1, len(net.Weights))
			continue
		}

		// 4. Check each neuron layer count
		for n, neuron := range net.Neurons {
			if uint(neuron.ColView(0).Len()) != test[n] {
				t.Errorf("Expected %d neurons on layer %d, got %d", test[n], n, neuron.ColView(0).Len())
			}
		}
		// 5. Check each bias layer count
		for b, bias := range net.Biases {

			if uint(bias.ColView(0).Len()) != test[b+1] {
				t.Errorf("Expected %d biases on layer %d, got %d", test[b+1], b, bias.ColView(0).Len())
			}

		}

		// 6. Check each weight layer count
		for w, weight := range net.Weights {

			rows, cols := weight.Dims()

			if uint(rows) != test[w+1] { // invalid rows
				t.Errorf("Expected %d weights' rows from layer %d to %d, got %d", test[w+1], w, w+1, rows)
			}
			if uint(cols) != test[w] { // invalid cols
				t.Errorf("Expected %d weights' cols from layer %d to %d, got %d", test[w], w, w+1, cols)
			}

		}

	}

}

func TestForwardPass(t *testing.T) {

	tests := []struct {
		Layers []uint
		X      []float64
	}{
		{
			[]uint{2, 2, 1},
			[]float64{0.5, 0.2},
		}, {
			[]uint{4, 5, 10},
			[]float64{0.5, 0.2, 0.8, 0.4},
		},
	}

	for _, test := range tests {

		net, err := Empty(test.Layers...)
		if err != nil {
			t.Errorf("Unexpected error <%s>", err)
			break
		}

		// apply forward pass
		_, err = net.Guess(test.X...)
		if err != nil {
			t.Errorf("Unexpected error <%s>", err)
			break
		}

		// check layer by layer (begin at 1)
		for l, ll := 1, len(net.layers); l < ll; l++ {

			// each neuron = ( each previous neuron times its weight ) + neuron bias
			for n, nl := 0, net.Neurons[l].ColView(0).Len(); n < nl; n++ {
				sum := net.Biases[l-1].At(n, 0)

				// sum each previous neuron*its weight
				for i, il := 0, net.Neurons[l-1].ColView(0).Len(); i < il; i++ {
					sum += net.Neurons[l-1].At(i, 0) * net.Weights[l-1].At(n, i)
				}

				sum = sigmoid(0, 0, sum)

				// check sum
				if !floats.EqualWithinAbs(net.Neurons[l].At(n, 0), sum, 1e9) {
					t.Fatalf("Expected neuron %d.%d to be %f, got %f", l, n, sum, net.Neurons[l].At(n, 0))
				}
			}

		}

	}
}