Towards Trust and Transparency in Deep Learning Systems through Behavior Introspection & Online Competency Prediction