Boosting - What is the theory behind boosting?

Iteratively combining weak learners is what boosting techniques do to generate a strong learner with better prediction capabilities. Weak learner models are those that do somewhat better than a random guess when it comes to data categorization.