Topic Modeling for Large-Scale Multimedia Analysis and Retrieval

The explosion of multimedia data in social media raises a great demand for developing effective and efficient computational tools to facilitate producing, analyzing, and retrieving large-scale multimedia content. Probabilistic topic models prove to be an effective way to organize large volumes of text documents, while much fewer related models are proposed for other types of unstructured data such as multimedia content, partly due to the high computational cost. With the emergence of cloud computing, topic models are expected to become increasingly applicable to multimedia data. Furthermore, the growing demand for a deep understanding of multimedia data on the web drives the development of sophisticated machine learning methods. Thus, it is greatly desirable to develop topic modeling approaches to multimedia applications that are consistently effective, highly efficient, and easily scalable. In this chapter, we present a review of topic models for large-scale multimedia analysis. Our goal is to show the current challenges from various perspectives and to present a Significant language editing was done. please check.