View source for Philgoo Han writeup of Cohen and Carvalho

This is a review of [[reviewed paper::Cohen_2005_stacked_sequential_learning]] by [[reviewer::user:Ironfoot]].

* Meta learning algorithm for any base learner
* High error rate for MEMM (for the particular training data)
** Different from label bias or observation bias. Due to too strong weight on history feature.
** <span style="color:navy">Since training and test data are from the same source the two should have same characterisitcs. One reason I can think of the great error rate is a single false prediction makes all the following text line flase in MEMM. But can this single reason magnify the error rate more than ten times bigger?</span>
* Sequential stacking
** Making a prediction of y
** <span style="color:navy">Why is "y^ similar to the prediction produced by an f learned by A on a size-m sample that does not include x"?</span>
** <span style="color:navy">I don't get the meaning of "f will not be used as the inner loop of a Viterbi or beam-search"</span>
* Result
** Much lower error rate. Using moderately large window(history) size improves the precision.
** <span style="color:navy">This may be an implementation issue. How do you handle the boundary conditions where there is not enough history or future states</span>
* Also work well on different data.
** It seems that sequential stacking algorithm can improve CMM bias problem in general.