| Scenario |
Type |
Detail |
| Steps (cycles) |
Basic Path |
The user chose a number of steps (cycles through given data) to halt training the network. During the step a parameter is iterated and then compared to halting parameter. When there is an equation, training ends. |
| Timeout |
Basic Path |
The user chose timeout for training the network, so when the time is greater than given halt parameter of miliseconds, training ends. Every step this parameter is compared to actual time. When the user trains the net by the steps, this parameter does not have a reason. |
| Total Error |
Basic Path |
It is possible to calculate total error of the net during every cycle. This error is going to decrease and that is why, the user can set error as stopping parameter. It is calculated in every step and then compared to given halting parameter. If the error is lower, training ends. |
|