*
* Learning rate should be a float value between 0.0(exclusive) and 1.0 (inclusive)
* Maximum number of iterations can be an integer value greater than 0
* If normalizeInputs is set to true, then every input given to the algorithm will be standardized
* by use of standard deviation and mean calculation
*
* @param int $learningRate
* @param int $maxIterations
*/
public function __construct(float $learningRate = 0.001, int $maxIterations = 1000,
bool $normalizeInputs = true, int $trainingType = self::BATCH_TRAINING)
{
if (! in_array($trainingType, [self::BATCH_TRAINING, self::ONLINE_TRAINING])) {
throw new \Exception("Adaline can only be trained with batch and online/stochastic gradient descent algorithm");
}
$this->trainingType = $trainingType;
parent::__construct($learningRate, $maxIterations, $normalizeInputs);
}
/**
* Adapts the weights with respect to given samples and targets
* by use of gradient descent learning rule
*/
protected function runTraining()
{
// The cost function is the sum of squares
$callback = function ($weights, $sample, $target) {
$this->weights = $weights;
$output = $this->output($sample);
$gradient = $output - $target;
$error = $gradient ** 2;
return [$error, $gradient];
};
$isBatch = $this->trainingType == self::BATCH_TRAINING;
return parent::runGradientDescent($callback, $isBatch);
}
}