In “Negative Correlation Learning for Classification Ensembles” Wang, Chen and Yao (2010) modify adaboost to put more weight on classifiers that are less correlated to the boosted classifier.  They do this by introducing a penalty term “
$$p_t = (h_t – H) \sum_{k\neq t} (h_k – H),$$
where H is the final output by combining the decisions from the individuals.”
The idea for this came from the Cooperative ensemble learning system (CELS) by Liu and Yao (1999).  Experimental results and performance bounds were given in “The Effectiveness of a New Negative Correlation Learning Algorithm for Classification Ensembles” Wang and Yao (2010).