How frequently does Matlab neural network toolbox functions updates the Weights? one training vector at a time? a full batch average (average of all vectors in each epoch)? or by mini batches? In the last case, what criteria is used to choose the batches' size?
K. Kaveh
Training can be implemented in two different ways: incremental and batch mode. MATLAB toolkit uses batch mode in which updating of the weights and biases is done after all inputs have been presented to the network. In the incremental mode, the weights and biases are updated after each input is applied to the network. Based on my experience batch mode provides better results than incremental.
2 votes
1 thanks
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Programming Languages
More Glen Dario Rodriguez's questions See All
Similar questions and discussions