Learning network representation has a variety of applications, such as network classification. Most existing work in this area focuses on static undirected networks and does not account for presence of directed edges or temporal changes. Furthermore, most work focuses on node representations that do poorly on tasks like network classification. In this paper, we propose a novel network embedding methodology, gl2vec, for network classification in both static and temporal directed networks. gl2vec constructs vectors for feature representation using static or temporal network graphlet distributions and a null model for comparing them against random graphs. We demonstrate the efficacy and usability of gl2vec over existing state-of-the-art methods on network classification tasks such as network type classification and subgraph identification in several real-world static and temporal directed networks. We argue that gl2vec provides additional network features that are not captured by state-of-the-art methods, which can significantly improve their classification accuracy by up to 10% in real-world applications.