Fingerprint is considered as a dominant biometric trait due to its acceptability, reliability, high security level and low cost. Due to the high demand on fingerprint identification system deployments, a lot of challenges are keep arising in each system's phase including fingerprint image enhancement, feature extraction, features matching and fingerprint classification. Machine learning techniques introduce non traditional solutions to the fingerprint identification challenges. This paper presents a short survey that emphasizes the implementations of basic machine learning notions for compensating some fingerprint problems. This survey contributes as a ground truth for developing machine learning based algorithms for fingerprint identification in the near future.