Hi Mahboobeh,

Thanks! I had seen that, but it doesn't concern it using classification accuracy as a scoring function. Her technical report that the presentation is based on is also very useful!

.

https://www.aaai.org/Papers/AAAI/2004/AAAI04-058.pdf

http://www.machinelearning.org/proceedings/icml2005/papers/083_Discriminative_PernkopfBilmes.pdf

optimize classification accuracy fpr structure learning

.

http://www.kevinjing.com/jing_icml.pdf

http://paperity.org/p/7462565/boosted-bayesian-network-classifiers

use accuracy to boost a BN ensemble

.

accuracy is also commonly used when using GA-like algorithms (as a "fitness" score), among other criteria :

http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.308.9725&rep=rep1&type=pdf