I have been working on ensemble learning and I came across this doubt that unlike other ensemble learning algorithms like voting classifier a can we only use one classifier with boosting.
Boosting typically only use one algorithm as it's base learner (almost exclusively decision trees). However, you could use a mixed set of algorithms as your base learners.
Something like this:
Boosting round 0: Add decision tree Boosting round 1: Add neural network Boosting round 2: Add KNN Boosting round 3: Add decision tree ...
The reason you only see boosting using the same algorithm is probably just because it works better. I speculate that the diversity that comes from using several algorithms shine more when they are trained in parallel and combined. In boosting the base learners are trained in a sequence.