Machine Learning for Audio, Image and Video Analysis: Theory and Applications (Advanced Information and Knowledge Processing)

Machine Learning involves several scientific domains including mathematics, computer science, statistics and biology, and is an approach that enables computers to automatically learn from data. Focusing on complex media and how to convert raw data into useful information, this book offers both introductory and advanced material in the combined fields of machine learning and image/video processing. The machine learning techniques presented enable readers to address many real world problems involving complex data. Examples covering areas such as automatic speech and handwriting transcription, automatic face recognition, and semantic video segmentation are included, along with detailed introductions to algorithms and examples of their applications. The book is organized in four parts: The first focuses on technical aspects, basic mathematical notions and elementary machine learning techniques. The second provides an extensive survey of most relevant machine learning techniques for media processing, while the third part focuses on applications and shows how techniques are applied in actual problems. The fourth part contains detailed appendices that provide notions about the main mathematical instruments used throughout the text. Students and researchers needing a solid foundation or reference, and practitioners interested in discovering more about the state-of-the-art will find this book invaluable. Examples and problems are based on data and software packages publicly available on the web. Front Matter....Pages I-XVI Introduction....Pages 1-10 Audio Acquisition, Representation and Storage....Pages 13-50 Image and Video Acquisition, Representation and Storage....Pages 51-80 Machine Learning....Pages 83-89 Bayesian Theory of Decision....Pages 91-115 Clustering Methods....Pages 117-148 Foundations of Statistical Learning and Model Selection....Pages 149-172 Supervised Neural Networks and Ensemble Methods....Pages 173-209 Kernel Methods....Pages 211-263 Markovian Models for Sequential Data....Pages 265-303 Feature Extraction Methods and Manifold Learning Methods....Pages 305-341 Speech and Handwriting Recognition....Pages 345-379 Automatic Face Recognition....Pages 381-411 Video Segmentation and Keyframe Extraction....Pages 413-430 Back Matter....Pages 431-494 Focusing on complex media and how to convert raw data into useful information, this book offers both introductory and advanced material in the combined fields of machine learning and image/video processing. It is organized into three parts. The first focuses on technical aspects, basic mathematical notions and elementary machine learning techniques. The second provides an extensive survey of most relevant machine learning techniques for media processing. The third focuses on applications and shows how techniques are applied in actual problems. Examples and problems are based on data and software packages publicly available on the web. This book illustrates how to deal with complex media and convert raw data into useful information. Students and researchers needing a solid foundation or reference, and practitioners interested in discovering more about the state-of-the-art will find this book invaluable. Illustrates how to deal with complex media and convert raw data into useful information. Once the original data have been converted into a digital representation, the processing of different media can be performed under the unifying framework of machine learning