Abstract:
In recent years, there has been a surge of interest in the application of machine learning techniques for media content processing. Challenges in complexity and variability of multimedia data have spawned many new developments in machine learning theories and algorithms. Applying machine learning techniques to multimedia content involves special considerations: the data is typically of very high dimension, the number of examples per class may be low, and the problem at hand does not always fit in one of the standard learning formulations. Furthermore, multimedia data, such as digital images, audio streams and motion video programs, exhibit a rich structured form, motivating the development of techniques to directly deal with structured information.
In this talk we will discuss some of these challenges, exemplifying with on-going research projects at the Telecommunications and Multimedia Unit of INESC Porto and trying to engage the audience for multimedia applications.