Bøger i Neural Information Processing series serien

An up-to-date account of the interplay between optimization and machine learning, accessible to students and researchers in both communities.The interplay between optimization and machine learning is one of the most important developments in modern computational science. Optimization formulations and methods are proving to be vital in designing algorithms to extract essential knowledge from huge volumes of data. Machine learning, however, is not simply a consumer of optimization technology but a rapidly evolving field that is itself generating new optimization ideas. This book captures the state of the art of the interaction between optimization and machine learning in a way that is accessible to researchers in both fields.Optimization approaches have enjoyed prominence in machine learning because of their wide applicability and attractive theoretical properties. The increasing complexity, size, and variety of today''s machine learning models call for the reassessment of existing assumptions. This book starts the process of reassessment. It describes the resurgence in novel contexts of established frameworks such as first-order methods, stochastic approximations, convex relaxations, interior-point methods, and proximal methods. It also devotes attention to newer themes such as regularized optimization, robust optimization, gradient and subgradient methods, splitting techniques, and second-order methods. Many of these techniques draw inspiration from other fields, including operations research, theoretical computer science, and subfields of optimization. The book will enrich the ongoing cross-fertilization between the machine learning community and these other fields, and within the broader optimization community.

Recent advances in the area of lifted inference, which exploits the structure inherent in relational probabilistic models.Statistical relational AI (StaRAI) studies the integration of reasoning under uncertainty with reasoning about individuals and relations. The representations used are often called relational probabilistic models. Lifted inference is about how to exploit the structure inherent in relational probabilistic models, either in the way they are expressed or by extracting structure from observations. This book covers recent significant advances in the area of lifted inference, providing a unifying introduction to this very active field. After providing necessary background on probabilistic graphical models, relational probabilistic models, and learning inside these models, the book turns to lifted inference, first covering exact inference and then approximate inference. In addition, the book considers the theory of liftability and acting in relational domains, which allows the connection of learning and reasoning in relational domains.

Advances in training models with log-linear structures, with topics including variable selection, the geometry of neural nets, and applications.Log-linear models play a key role in modern big data and machine learning applications. From simple binary classification models through partition functions, conditional random fields, and neural nets, log-linear structure is closely related to performance in certain applications and influences fitting techniques used to train models. This volume covers recent advances in training models with log-linear structures, covering the underlying geometry, optimization techniques, and multiple applications. The first chapter shows readers the inner workings of machine learning, providing insights into the geometry of log-linear and neural net models. The other chapters range from introductory material to optimization techniques to involved use cases. The book, which grew out of a NIPS workshop, is suitable for graduate students doing research in machine learning, in particular deep learning, variable selection, and applications to speech recognition. The contributors come from academia and industry, allowing readers to view the field from both perspectives.ContributorsAleksandr Aravkin, Avishy Carmi, Guillermo A. Cecchi, Anna Choromanska, Li Deng, Xinwei Deng, Jean Honorio, Tony Jebara, Huijing Jiang, Dimitri Kanevsky, Brian Kingsbury, Fabrice Lambert, Aurélie C. Lozano, Daniel Moskovich, Yuriy S. Polyakov, Bhuvana Ramabhadran, Irina Rish, Dimitris Samaras, Tara N. Sainath, Hagen Soltau, Serge F. Timashev, Ewout van den Berg

State-of-the-art algorithms and theory in a novel domain of machine learning, prediction when the output has structure.

A survey of probabilistic approaches to modeling and understanding brain function.

An overview of recent efforts in the machine learning community to deal with dataset and covariate shift, which occurs when test and training inputs and outputs have different distributions.