Support vector machines (SVMs) are one of the most active research areas in machine learning. For SVMs to gain a wider degree of acceptance in fields such as medical diagnosis and security sensitive areas, it is desirable to offer an explanation capability.