Conventional refers to any classifier that depend only on one algorithm to predict the class for ex. SVM, Knn, Naive. On the contrary, when two or more classifiers are hybridized together to exploit the advantages of each, such classier is referred to as non-conventional as it depend on more than one algorithm to predict the class.
Yes, so you did answered your Own question. Hybrid is a common way nowadays so as to exploit advantages of each of the used classifiers. so it's better to use Hybrid classifiers