We are witnessing an explosive proliferation of digital videos and images from a variety of sources such as the World Wide Web. Billions of images and videos are now freely available online (e.g. from video/image-sharing sites YouTube and Flickr), which constitute a dense sampling of the visual world. Such huge volumes of multimedia data pose many critical challenges for large-scale multimedia mining applications such as semantic category discovery, user behavior analysis, and trend (e.g., hot topic) prediction. Existing multimedia mining technology is still in its premature stage and very often cannot meet the requirements of real life users such as real-time monitoring or retrieval of large-scale multimedia collections. The increasing demand for visual content analysis and retrieval of large-scale visual data has attracted extensive research efforts in the multimedia community.

To meet these challenges, this workshop is soliciting research papers that investigate novel theories, algorithms, or applications for large-scale multimedia mining. It aims to provide a forum for researchers to exchange ideas and demonstrate the-state-of-the-art research outputs in this emerging field, as well as to set up a platform for the industry to acquire and comment to the recent research developments. Innovative works with solid experimental validations over realistic large-scale data sets are particularly welcome.