Preface Statistical Learning Theory now plays a more active role:after the general analysis of learning processes,the res earch in the area of synthesis of optimal algorithms was started.These studies,however,do not belong to history yet.They are a sulject of today's research activities. Vladimir Vapnik (1995) The Support Vector Machine has recently been introduced as a new technique for solving a variety of learning and function estimation problems.During a workshop at the annual Neur al Information Processing Systems (NIPS)conference,held in Breckenridge,Colorado in December 1997,a snapshot of the state of the art in Support Vector learning was recorded.A variety of people helped in this,among them our co-organizer Leon Bottou,the NIPS workshop chairs Steve Nowlan and Rich Zemel,and all the workshop speakers and attendees who contributed to lively discussions.After the workshop,we decided that it would be worthwhile to invest some time to have the snapshot printed. We invited all the speakers as well as other researchers to submit papers for this collection,and integr ated the results into the present book.We believe that it covers the full range of current Support Vector research at an early point in time.This is possible for two reasons.First,the field of SV learning is in its early (and thus exciting)days.Second,this book gathers expertise from all contributers,whom we wholeheartedly thank for all the work they have put into our joint effort.Any single person trying to accomplish this task would most likely have failed:either by writing a book which is less comprehensive,or by taking more time to complete the book. It is our hope that this outweighs the shortcomings of the book,most notably the fact that a collection of chapters can never be as homogeneous as a book conceived by a single person.We have tried to compensate for this by the selection and refereeing process of the submissions.In addition,we have written an introductory chapter describing the SV algorithm in some detail (chapter 1),and added a roadmap (chapter 2)which describes the actual contributions which are to follow in chapters 3 through 20. Bernhar d Scholkopf,Christopher J.C.Burges,Alexander J.Smola Berlin,Holmdel,July 1998 1998/08/251631 Preface Statistical Learning Theory now plays a more active role after the general analysis of learning processes the research in the area of synthesis of optimal algorithms was started These studies however do not belong to history yet They are a subject of todays research activities Vladimir Vapnik The Support Vector Machine has recently been introduced as a new technique for solving a variety of learning and function estimation problems During a workshop at the annual Neural Information Processing Systems NIPS conference held in Breckenridge Colorado in December a snapshot of the state of the art in Support Vector learning was recorded A variety of people helped in this among them our coorganizer Leon Bottou the NIPS workshop chairs Steve Nowlan and Rich Zemel and all the workshop speakers and attendees who contributed to lively discussions After the workshop we decided that it would be worthwhile to invest some time to have the snapshot printed We invited all the speakers as well as other researchers to submit papers for this collection and integrated the results into the present book We believe that it covers the full range of current Support Vector research at an early point in time This is possible for two reasons First the eld of SV learning is in its early and thus exciting days Second this book gathers expertise from all contributers whom we wholeheartedly thank for all the work they have put into our joint e ort Any single person trying to accomplish this task would most likely have failed either by writing a book which is less comprehensive or by taking more time to complete the book It is our hope that this outweighs the shortcomings of the book most notably the fact that a collection of chapters can never be as homogeneous as a book conceived by a single person We have tried to compensate for this by the selection and refereeing process of the submissions In addition we have written an introductory chapter describing the SV algorithm in some detail chapter and added a roadmap chapter which describes the actual contributions which are to follow in chapters through Bernhard Scholkopf Christopher JC Burges Alexander J Smola Berlin Holmdel July