A "Bayes" classifier <http://en.wikipedia.org/wiki/Bayes_classifier> is a well defined entity that is semantically separate from the "Optimal Bayesian" Classifier. But you're right that the risk of confusion is probably low seeing as the Bayes classifier is primarily theoretical (given that the needed exact feature-label distribution is typically unknown).
Though now you've started me wondering about two more of my hastily named (and even more nascent) Julia libraries: DAI.jl and MCBN.jl ;).Though their full names are not very friendly either... (DiscreteApproximateInference.jl and MonteCarloBayesianNetworks.jl) Jason I suspect the optimal is necessary as it's a common rhetoric in certain > statistical communities. Cf. optimal classification as described in > political science. > > -- John > > On Aug 13, 2014, at 1:46 PM, Stefan Karpinski <[email protected]> > wrote: > > Cool stuff. May I preemptively suggest calling it BayesianClassifiers? I'm > assuming the "optimal" part is redundant since no one will be clamoring for > suboptimal Bayesian classifiers. > >
