Few-Shot Concept Unlearning with Low Rank Adaptation