The inputSize of each layer does not need to be specified anymore
This commit is contained in:
parent
95501bf4b1
commit
1c66f1b72f
3 changed files with 60 additions and 41 deletions
|
@ -12,9 +12,9 @@ public class ExampleXOR {
|
|||
new SimpleMatrix(new double[][]{{0}})};
|
||||
|
||||
Network network = new Network();
|
||||
network.addLayer(new FCLayer(2, 3));
|
||||
network.addLayer(new FCLayer(3));
|
||||
network.addLayer(new ActivationLayer(ActivationFunctions::tanh, ActivationFunctions::tanhPrime));
|
||||
network.addLayer(new FCLayer(3, 1));
|
||||
network.addLayer(new FCLayer(1));
|
||||
network.addLayer(new ActivationLayer(ActivationFunctions::tanh, ActivationFunctions::tanhPrime));
|
||||
|
||||
network.use(LossFunctions::MSE, LossFunctions::MSEPrime);
|
||||
|
|
|
@ -12,9 +12,9 @@ public class ExampleXORBlankLayers {
|
|||
new SimpleMatrix(new double[][]{{0}})};
|
||||
|
||||
Network network = new Network();
|
||||
network.addLayer(new FCLayer(2, 1));
|
||||
network.addLayer(new FCLayer(1));
|
||||
network.addLayer(new ActivationLayer(ActivationFunctions::tanh, ActivationFunctions::tanhPrime));
|
||||
network.addLayer(new FCLayer(1, 1));
|
||||
network.addLayer(new FCLayer(1));
|
||||
network.addLayer(new ActivationLayer(ActivationFunctions::tanh, ActivationFunctions::tanhPrime));
|
||||
network.addNeuron(0, 2);
|
||||
|
||||
|
|
|
@ -3,15 +3,23 @@ import org.ejml.simple.SimpleMatrix;
|
|||
import java.util.Random;
|
||||
|
||||
public class FCLayer extends Layer {
|
||||
SimpleMatrix weights;
|
||||
SimpleMatrix biases;
|
||||
private SimpleMatrix weights;
|
||||
private SimpleMatrix biases;
|
||||
private int numNeurons;
|
||||
private boolean isInitialized;
|
||||
|
||||
public FCLayer(int inputSize, int outputSize) {
|
||||
public FCLayer(int numNeurons) {
|
||||
this.numNeurons = numNeurons;
|
||||
isInitialized = false;
|
||||
}
|
||||
|
||||
private void initialize(int inputSize) {
|
||||
Random random = new Random();
|
||||
weights = new SimpleMatrix(inputSize, outputSize, true,
|
||||
random.doubles((long) inputSize*outputSize, -1, 1).toArray());
|
||||
biases = new SimpleMatrix(1, outputSize, true,
|
||||
random.doubles(outputSize, -1, 1).toArray());
|
||||
this.weights = new SimpleMatrix(inputSize, numNeurons, true,
|
||||
random.doubles((long) inputSize*numNeurons, -1, 1).toArray());
|
||||
this.biases = new SimpleMatrix(1, numNeurons, true,
|
||||
random.doubles(numNeurons, -1, 1).toArray());
|
||||
this.isInitialized = true;
|
||||
}
|
||||
|
||||
/**
|
||||
|
@ -19,6 +27,7 @@ public class FCLayer extends Layer {
|
|||
* @param n amount of new neurons in previous layer
|
||||
*/
|
||||
public void updateInputSize(int n) {
|
||||
if (isInitialized) {
|
||||
Random random = new Random();
|
||||
|
||||
// add new weights
|
||||
|
@ -35,6 +44,7 @@ public class FCLayer extends Layer {
|
|||
}
|
||||
this.weights = newWeights;
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Adds new neurons at the end of the layer
|
||||
|
@ -43,6 +53,10 @@ public class FCLayer extends Layer {
|
|||
public void addNeuron(int n) {
|
||||
Random random = new Random();
|
||||
|
||||
// update neuron count
|
||||
this.numNeurons += n;
|
||||
|
||||
if (isInitialized) {
|
||||
// add new weights
|
||||
SimpleMatrix newWeights = new SimpleMatrix(this.weights.numRows(), this.weights.numCols() + n);
|
||||
for (int i = 0; i < this.weights.numRows(); i++) {
|
||||
|
@ -64,9 +78,14 @@ public class FCLayer extends Layer {
|
|||
System.arraycopy(newBiasValues, 0, newBiases.getDDRM().data, this.biases.numCols(), n);
|
||||
this.biases = newBiases;
|
||||
}
|
||||
}
|
||||
|
||||
@Override
|
||||
public SimpleMatrix forwardPropagation(SimpleMatrix inputs) {
|
||||
if (!isInitialized) {
|
||||
initialize(inputs.numCols());
|
||||
}
|
||||
|
||||
this.input = inputs;
|
||||
this.output = this.input.mult(this.weights).plus(this.biases);
|
||||
return this.output;
|
||||
|
|
Loading…
Reference in a new issue