Which samples should be labelled in a large data set is one of the most important problems for training of deep learning. So far, a variety of active sample selection strategies related to deep learning have been proposed in many literatures. We defined them as Active Deep Learning (ADL) only if their predictor is deep model, where the basic learner is called as predictor and the labeling schemes is called selector. In this survey, three fundamental factors in selector designation were summarized. We category ADL into model-driven ADL and data-driven ADL, by whether its selector is model-driven or data-driven. The different characteristics of the two major type of ADL were addressed in detail respectively. Furthermore, different sub-classes of data-driven and model-driven ADL are also summarized and discussed emphatically. The advantages and disadvantages between data-driven ADLand model-driven ADL are thoroughly analyzed. We pointed out that, with the development of deep learning, the selector in ADL also is experiencing the stage from model-driven to data-driven. Finally,we make discussion on ADL about its uncertainty, explanatory, foundations of cognitive science etc.and survey on the trend of ADL from model-driven to data-driven.
Aerospace Information Research Institute (AIR), Chinese Academy of Sciences (CAS)