Multi Layer Perceptron#

A multiclass feedforward neural network classifier with user-defined hidden layers as intermediate computational units. Multiple layers and non-linear activation functions allow the Multi Layer Perceptron to handle complex non-linear problems. In addition, the MLP features progress monitoring which stops training when it can no longer make progress. It also utilizes network snapshotting to make sure that it always has the best model parameters even if progress declined during training.

Interfaces: Estimator, Learner, Online, Probabilistic, Verbose, Persistable

Data Type Compatibility: Continuous


# Param Default Type Description
1 hidden array An array composing the user-specified hidden layers of the network in order.
2 batch size 100 int The number of training samples to process at a time.
3 optimizer Adam object The gradient descent optimizer used to update the network parameters.
4 alpha 1e-4 float The amount of L2 regularization to apply to the parameters of the network.
5 epochs 1000 int The maximum number of training epochs. i.e. the number of times to iterate over the entire training set before terminating.
6 min change 1e-4 float The minimum change in the training loss necessary to continue training.
7 window 3 int The number of epochs without improvement in the validation score to wait before considering an early stop.
8 holdout 0.1 float The proportion of training samples to use for validation and progress monitoring.
9 cost fn CrossEntropy object The function that computes the loss associated with an erroneous activation during training.
10 metric FBeta object The validation metric used to score the generalization performance of the model during training.

Additional Methods#

Return the training loss at each epoch:

public steps() : array

Return the validation scores at each epoch:

public scores() : array

Returns the underlying neural network instance or null if untrained:

public network() : Network|null


use Rubix\ML\Classifiers\MultiLayerPerceptron;
use Rubix\ML\NeuralNet\Layers\Dense;
use Rubix\ML\NeuralNet\Layers\Dropout;
use Rubix\ML\NeuralNet\Layers\Activation;
use Rubix\ML\NeuralNet\ActivationFunctions\LeakyReLU;
use Rubix\ML\NeuralNet\ActivationFunctions\PReLU;
use Rubix\ML\NeuralNet\Optimizers\Adam;
use Rubix\ML\NeuralNet\CostFunctions\CrossEntropy;
use Rubix\ML\CrossValidation\Metrics\MCC;

$estimator = new MultiLayerPerceptron([
    new Dense(200),
    new Activation(new LeakyReLU()),
    new Dropout(0.3),
    new Dense(100),
    new Activation(new LeakyReLU()),
    new Dropout(0.3),
    new Dense(50),
    new PReLU(),
], 100, new Adam(0.001), 1e-4, 1000, 1e-3, 3, 0.1, new CrossEntropy(), new MCC());


  • G. E. Hinton. (1989). Connectionist learning procedures.
  • L. Prechelt. (1997). Early Stopping - but when?