Transfer learning is a machine learning technique where a model uses knowledge gained from previously learned tasks and applies it to new, related tasks or activities.