división de datos en conjuntos de validación de pruebas de entrenamiento en Matlab 2010

my question is: Matlab 2010 provides options of Testing, Validation periods in Neural Network process. is this data splitting or will i have to use "crossvalind" for data splitting?

preguntado el 04 de julio de 12 a las 08:07

1 Respuestas

Aquí hay un extracto del documentación:

When training multilayer networks, the general practice is to first divide the data into three subsets. The first subset is the training set, which is used for computing the gradient and updating the network weights and biases. The second subset is the validation set. The error on the validation set is monitored during the training process. [...] The test set error is not used during training, but it is used to compare different models. [...]

There are four functions provided for dividing data into training, validation and test sets: dividerand, divideblock, divideint y divideind. (actually there is a fifth dividetrain that assigns all instances to training)

mesa

For more sophisticated methods (cross-validation, stratification, etc..), check out cvpartition or crossvalind funciones.

Respondido 04 Jul 12, 09:07

No es la respuesta que estás buscando? Examinar otras preguntas etiquetadas or haz tu propia pregunta.