Method for determination of weights, suitable for elimination, of a neural network using a computer

A method for determining weights that are for removal in a neural network, using a computer, comprising the steps of:

• stopping a training phase of the neural network before an error function, which is to be minimized in the training phase, reaches a minimum;
• defining a first variable for at least one weight of the neural network, which first variable is used to describe an assessment of the at least one weight in terms of removal of the at least one weight from the neural network, based on the error function being at the minimum;
• defining a second variable for the at least one weight, which second variable is used to describe an extent to which a value of the error function varies when the at least one weight varies;
• determining a criterion variable for the at least one weight from at least the first variable and the second variable, which criterion variable is used to describe the extent to which the value of the error function varies if the at least one weight is removed from the neural network;
• classifying the at least one weight, as a weight that is suitable for removal, if the criterion variable results in the at least one weight being such that the removal of the at least one weight varies the value of the error function by a value which is less than a first limit that is predeterminable;
• and
• removing the at least one weight that is suitable for removal from the neural network.