Learn the first story in depth, sprinkle flowers ~ ~ ~
Contact machine learning for nearly a year, contact depth study for three or four months, feel that each training code is just no brain training, this time first to tidy up about episode and batch related knowledge.
Epoch is the number of training sessions for all training data.
A small part of the batch used in one training
The specific relationship is as follows: (pseudo code from here?)
Loop Maxepochs Times
For each training item
compute weights and bias deltas for Curr item
accumulate the deltas end for
Adjust Weights and bias deltas using accumulated deltas end
loop
For online training, the pseudocode is as follows: (or from there?)
Loop maxepochs times for each
training the data item
compute weights and bias deltas for Curr item
adjust weights and bias values using deltas end to end
loop
(It seems like it's going to end ...) = =)