1. Memetic Algorithms are Evolutive Metaheuristics based on Local Search, contrary to general Genetic Algorithms. In any case, both refer to population-based methods aiming at "evolving" certain qualities throughout the iterative process. This can indeed be interpreted as following a learning strategy (not to confuse with adaptive methods).
2. Matheuristics are, as Pr Letchford indicated, heuristic algorithms using exact algorithms to solve (usually) MILP models, either on a one-shot or iterative basis. Hence, a Matheuristic using a Memetic Algorithm would somehow comprise a "learning strategy" but they are not to be confused.
An excellent reference on Matheuristics (for routing problems) is that of Archetti and Speranza 2014. Hope you find it useful.
As Adam said, it would qualify as a matheuristic, although this may depend on how you precisely define the latter. Note that you can use exact methods not just as local improvers, but in other parts of a MA, such as for example in the recombination operator. This kind of combinations have been around since the mid 90s. Please, check the attached references.
Conference Paper Combining Metaheuristics and Exact Algorithms in Combinatori...
Chapter Memetic Algorithms and Complete Techniques