Multimedia information such as texts, sounds, images, videos, are ubiquitous in modern society. In computer science, an extremely large number of real-world tasks. For instance, image matching/registration/retrieval, object detection, dimensionality reduction, and subspace clustering are aimed at discovering patterns or structures in observed data, which is the key to understanding and processing such data.

In the last two decades, a variety of local and global patterns/structures (e.g., graphical and geometric priors, low rankness, sparsity, exclusivity, and mutuality) have been designed and widely applied. Though existing techniques have shown their efficacy, more advanced and sophisticated development is needed to handle more complex data and aid more real-world tasks. Furthermore, with the emergence of deep learning models, there is an expectation of exploiting more powerful tools for practical use.

This Special Issue aims to solicit contributions on the advanced pattern/structure discovery from complex multimedia data environments. Original research articles presenting novel in-depth fundamental research are welcome, along with review articles discussing the current state-of-the-art.

Potential topics include but are not limited to the following:

• Classification, clustering, and collaborative filtering
• Deep feature learning and feature selection
• Dimensionality reduction and manifold learning
• Low rank matrix/tensor recovery, linear/nonlinear regression, and sparse coding
• Multi-task and transfer learning, online learning, and metric learning
• Reinforcement learning and representation learning
• Deconvolution/deblurring, denoising, and image restoration
• Image/feature matching/registration and multimedia retrieval
• Object detection and recognition, motion and tracking
• Image fusion, image super-resolution, and image segmentation