This project seeks to perform a thorough empirical study of the sample and computational complexity and learning effectiveness of now popular deep-learning techniques and a fair comparison with traditiona/standard machine-learning techniques that controls for available computational resources.