Machine Learning Techniques for Multimedia Analysis
Slim Essid, Marine Campedel, Gaël Richard, Tomáš Piatrik, Rachid Benmokhtar, Benoît Huet · 2011
Humans have natural abilities to categorize objects, images or sounds and to place them in specific classes in which they will share some common characteristics, patterns or semantics. More generally, the fundamental task of a classification algorithm is therefore to put data into groups (or classes) according to similarity measures with or without a priori knowledge about the classes considered. The approaches are termed supervised when a priori information (such as the labels) is used. They are otherwise called unsupervised. The aim of this chapter is not to give an exhaustive overview of existing machine learning techniques but rather to present some of the main approaches for setting up an automatic multimedia analysis system. It is dedicated to feature dimensionality reduction by automatic feature selection algorithms. The chapter discusses the classification approaches including both supervised and unsupervised techniques. It discusses the fusion (or consensus) approaches. Controlled Vocabulary Terms learning (artificial intelligence); multimedia systems; semantic networks; unsupervised learning