Statistical Machine Learning vs Deep Learning in Information Fusion: Competition or Collaboration?

摘要

Information fusion is the process of coherently and intelligently combining knowledge extracted from different sensors/modalities, in order to obtain more useful or discriminant information for the purpose of multimedia processing and biometrics, among others. The key to successful information fusion is to intelligently exploit the intrinsic relations between the data of different modalities. Statistical machine learning (SML) has played a major role in developing new information fusion methods, by incorporating prior knowledge and entropy metric, correlation analysis, inherent statistical structures of input data, and nonlinear relations. On the other hand, the recent development of deep learning (DL) draws enormous attention from the machine learning community. DL algorithms possess deep structures, requiring a large amount of data to train the huge number of parameters, an ultra-expensive process. However, the payoff is enormous; unprecedented success in many applications. This paper will first review recent development of both SML and DL in the context of information fusion, then analyze their pros and cons, and compare their performance in a number of application domains. Based on preliminary results, some thoughts will be presented on how SML and DL can work together to bring the study in machine learning to the next level, better serving human needs.