Stacked sequential learning write up

This is a review of Cohen_2005_stacked_sequential_learning by user:sgopal1

• The complexity would still be O(K). Assuming that K is not very small ( in that case the number of features to Naive Bayes will increase linearly with K ) it will approximately incur 10 times more cost to train. This is excluding the time taken to build the original classifier without augmenting the label features.