This represents a complete convergence of the descent, but can take a long time to occur, as the update for each weight must become smaller than can be represented in floating points.
Inheritance: implements MCordingley\Regression\Algorithm\GradientDescent\StoppingCriteria\StoppingCriteria
CoefficientEquality Class Documentation
Beispiel #1
0
 public function testConverged()
 {
     $criteria = new CoefficientEquality();
     static::assertFalse($criteria->converged([], [1, 2]));
     static::assertFalse($criteria->converged([], [1, 2, 3]));
     static::assertTrue($criteria->converged([], [1, 2, 3]));
 }