Machinelearningis arapidlymaturing?eldthataims toprovidepracticalme- ods for data discovery, categorization and modelling. The She?eld Machine Learning Workshop, which was held 7-10 September 2004, brought together some of the leading international researchers in the ?eld for a series of talks and posters that represented new developments in machine learning and numerical methods. The workshop was sponsored by the Engineering and Physical Sciences - search Council (EPSRC) and the London Mathematical Society (LMS) through the MathFIT program,whose aim is the encouragementof new interdisciplinary research.AdditionalfundingwasprovidedbythePASCALEuropeanFramework 6 Network of Excellence and the University of She?eld. It was the commitment of these funding bodies that enabled the workshop to have a strong program of invited speakers,and the organizerswish to thank these funding bodies for their ?nancial support. The particular focus for interactions at the workshop was - vanced Research Methods in Machine Learning and Statistical Signal Processing. These proceedings contain work that was presented at the workshop, and ideas that were developed through, or inspired by, attendance at the workshop. The proceedings re?ect this mixture and illustrate the diversity of applications and theoretical work in machine learning. We would like to thank the presenters and attendees at the workshop for the excellent quality of presentation and discussion during the oral and poster sessions. We are also grateful to Gillian Callaghan for her support in the orga- zation of the workshop, and ?nally we wish to thank the anonymous reviewers for their help in compiling the proceedings.

Genetic algorithms (GAs) are powerful search techniques based on principles of evolution and widely applied to solve problems in many disciplines. However, most GAs employed in practice nowadays are unable to learn genetic linkage and suffer from the linkage problem. The linkage learning genetic algorithm (LLGA) was proposed to tackle the linkage problem with several specially designed mechanisms. While the LLGA performs much better on badly scaled problems than simple GAs, it does not work well on uniformly scaled problems as other competent GAs. Therefore, we need to understand why it is so and need to know how to design a better LLGA or whether there are certain limits of such a linkage learning process. This book aims to gain better understanding of the LLGA in theory and to improve the LLGA's performance in practice. It starts with a survey of the existing genetic linkage learning techniques and describes the steps and approaches taken to tackle the research topics, including using promoters, developing the convergence time model, and adopting subchromosomes.

"Machine Learning and Data Mining for Computer Security" provides an overview of the current state of research in machine learning and data mining as it applies to problems in computer security. This book has a strong focus on information processing and combines and extends results from computer security.

The first part of the book surveys the data sources, the learning and mining methods, evaluation methodologies, and past work relevant for computer security. The second part of the book consists of articles written by the top researchers working in this area. These articles deals with topics of host-based intrusion detection through the analysis of audit trails, of command sequences and of system calls as well as network intrusion detection through the analysis of TCP packets and the detection of malicious executables.

This book fills the great need for a book that collects and frames work on developing and applying methods from machine learning and data mining to problems in computer security.

This volume contains the papers presented at the 16th Annual International Conference on Algorithmic Learning Theory (ALT 2005), which was held in S- gapore (Republic of Singapore), October 8-11, 2005. The main objective of the conference is to provide an interdisciplinary forum for the discussion of the t- oretical foundations of machine learning as well as their relevance to practical applications. The conference was co-located with the 8th International Conf- enceonDiscoveryScience(DS2005). Theconferencewasalsoheldinconjunction with the centennial celebrations of the National University of Singapore. The volume includes 30 technical contributions, which were selected by the program committee from 98 submissions. It also contains the ALT 2005 invited talks presented by Chih-Jen Lin (National Taiwan University, Taipei, Taiwan) on "Training Support Vector Machines via SMO-type Decomposition Methods," and by Vasant Honavar (Iowa State University, Ames, Iowa, USA) on "Al- rithmsandSoftwareforCollaborativeDiscoveryfromAutonomous, Semantically Heterogeneous, Distributed, Information Sources. " Furthermore, this volume - cludes an abstract of the joint invited talk with DS 2005 presented by Gary L. Bradshaw (Mississippi State University, Starkville, USA) on "Invention and Arti?cial Intelligence," and abstracts of the invited talks for DS 2005 presented by Ross D. King (The University of Wales, Aberystwyth, UK) on "The Robot Scientist Project," and by Neil Smalheiser (University of Illinois at Chicago, Chicago, USA) on "The Arrowsmith Project: 2005 Status Report. " The c- plete versions of these papers are published in the DS 2005 proceedings (Lecture Notes in Computer Science Vol. 3735).

The European Conference on Machine Learning (ECML) and the European Conference on Principles and Practice of Knowledge Discovery in Databases (PKDD) were jointly organized this year for the ?fth time in a row, after some years of mutual independence before. After Freiburg (2001), Helsinki (2002), Cavtat (2003) and Pisa (2004), Porto received the 16th edition of ECML and the 9th PKDD in October 3-7. Having the two conferences together seems to be working well: 585 di?erent paper submissions were received for both events, which maintains the high s- mission standard of last year. Of these, 335 were submitted to ECML only, 220 to PKDD only and 30 to both. Such a high volume of scienti?c work required a tremendous e?ort from Area Chairs, Program Committee members and some additional reviewers. On average, PC members had 10 papers to evaluate, and Area Chairs had 25 papers to decide upon. We managed to have 3 highly qua- ?edindependentreviewsperpaper(withveryfewexceptions)andoneadditional overall input from one of the Area Chairs. After the authors' responses and the online discussions for many of the papers, we arrived at the ?nal selection of 40 regular papers for ECML and 35 for PKDD. Besides these, 32 others were accepted as short papers for ECML and 35 for PKDD. This represents a joint acceptance rate of around 13% for regular papers and 25% overall. We thank all involved for all the e?ort with reviewing and selection of papers. Besidesthecoretechnicalprogram, ECMLandPKDDhad6invitedspeakers, 10 workshops, 8 tutorials and a Knowledge Discovery Challenge.

'A manual for the 21st-century citizen... accessible, refreshingly critical, relevant and urgent' - Financial Times 'Fascinating and deeply disturbing' - Yuval Noah Harari, Guardian Books of the Year In this New York Times bestseller, Cathy O'Neil, one of the first champions of algorithmic accountability, sounds an alarm on the mathematical models that pervade modern life -- and threaten to rip apart our social fabric. We live in the age of the algorithm. Increasingly, the decisions that affect our lives - where we go to school, whether we get a loan, how much we pay for insurance - are being made not by humans, but by mathematical models. In theory, this should lead to greater fairness: everyone is judged according to the same rules, and bias is eliminated. And yet, as Cathy O'Neil reveals in this urgent and necessary book, the opposite is true. The models being used today are opaque, unregulated, and incontestable, even when they're wrong. Most troubling, they reinforce discrimination. Tracing the arc of a person's life, O'Neil exposes the black box models that shape our future, both as individuals and as a society. These "weapons of math destruction" score teachers and students, sort CVs, grant or deny loans, evaluate workers, target voters, and monitor our health. O'Neil calls on modellers to take more responsibility for their algorithms and on policy makers to regulate their use. But in the end, it's up to us to become more savvy about the models that govern our lives. This important book empowers us to ask the tough questions, uncover the truth, and demand change.

This volume contains papers presented at the Eighteenth Annual Conference on Learning Theory (previously known as the Conference on Computational Learning Theory) held in Bertinoro, Italy from June 27 to 30, 2005. The technical program contained 45 papers selected from 120 submissions, 3 open problems selected from among 5 contributed, and 2 invited lectures. The invited lectures were given by Sergiu Hart on "Uncoupled Dynamics and Nash Equilibrium", and by Satinder Singh on "Rethinking State, Action, and Reward in Reinforcement Learning". These papers were not included in this volume. The Mark Fulk Award is presented annually for the best paper co-authored by a student. The student selected this year was Hadi Salmasian for the paper titled "The Spectral Method for General Mixture Models" co-authored with Ravindran Kannan and Santosh Vempala. The number of papers submitted to COLT this year was exceptionally high. In addition to the classical COLT topics, we found an increase in the number of submissions related to novel classi?cation scenarios such as ranking. This - crease re?ects a healthy shift towards more structured classi?cation problems, which are becoming increasingly relevant to practitioners.

Volume III of the Transactions on Rough Sets (TRS) introduces advances in the theory and application of rough sets. These advances have far-reaching impli- tions in a number of researchareas such as approximate reasoning, bioinform- ics, computerscience, datamining, engineering(especially, computerengineering and signal analysis), intelligent systems, knowledge discovery, pattern recog- tion, machineintelligence, andvariousformsoflearning. This volumerevealsthe vigor, breadth and depth in research either directly or indirectly related to the rough sets theory introduced by Prof. Zdzis law Pawlak more than three decades ago. Evidence of this can be found in the seminal paper on data mining by Prof. Pawlak included in this volume. In addition, there are eight papers on the theory and application of rough sets as well as a presentation of a new version of the Rough Set Exploration System (RSES) tool set and an introduction to the Rough Set Database System (RSDS). Prof. Pawlak has contributed a pioneering paper on data mining to this v- ume. In this paper, it is shown that information ?ow in a ?ow graph is governed by Bayes' rule with a deterministic rather than a probabilistic interpretation. A cardinal feature of this paper is that it is self-contained inasmuch as it not only introduces a new viewof information?owbut alsoprovidesanintroduction to the basic concepts of ?ow graphs. The representation of information ?ow - troduced in this paper makes it possible to study di?erent relationships in data and establishes a basis for a new mathematical tool for data mining. Inadditionto thepaperbyProf

Algorithmic Learning in a Random World describes recent theoretical and experimental developments in building computable approximations to Kolmogorov's algorithmic notion of randomness. Based on these approximations, a new set of machine learning algorithms have been developed that can be used to make predictions and to estimate their confidence and credibility in high-dimensional spaces under the usual assumption that the data are independent and identically distributed (assumption of randomness). Another aim of this unique monograph is to outline some limits of predictions: The approach based on algorithmic theory of randomness allows for the proof of impossibility of prediction in certain situations. The book describes how several important machine learning problems, such as density estimation in high-dimensional spaces, cannot be solved if the only assumption is randomness.

Get up and running with the basics of geographic information systems (GIS), geospatial analysis, and machine learning on spatial data in Python. This book starts with an introduction to geodata and covers topics such as GIS and common tools, standard formats of geographical data, and an overview of Python tools for geodata. Specifics and difficulties one may encounter when using geographical data are discussed: from coordinate systems and map projections to different geodata formats and types such as points, lines, polygons, and rasters. Analytics operations typically applied to geodata are explained such as clipping, intersecting, buffering, merging, dissolving, and erasing, with implementations in Python. Use cases and examples are included. The book also focuses on applying more advanced machine learning approaches to geographical data and presents interpolation, classification, regression, and clustering via examples and use cases. This book is your go-to resource for machine learning on geodata. It presents the basics of working with spatial data and advanced applications. Examples are presented using code (accessible at github.com/Apress/machine-learning-geographic-data-python) and facilitate learning by application. What You Will Learn Understand the fundamental concepts of working with geodata Work with multiple geographical data types and file formats in Python Create maps in Python Apply machine learning on geographical data Who This Book Is For Readers with a basic understanding of machine learning who wish to extend their skill set to analysis of and machine learning on spatial data while remaining in a common data science Python environment

This new edited volume consists of a collection of original articles written by leading financial economists and industry experts in the area of machine learning for asset management. The chapters introduce the reader to some of the latest research developments in the area of equity, multi-asset and factor investing. Each chapter deals with new methods for return and risk forecasting, stock selection, portfolio construction, performance attribution and transaction costs modeling. This volume will be of great help to portfolio managers, asset owners and consultants, as well as academics and students who want to improve their knowledge of machine learning in asset management.

Discover this multi-disciplinary and insightful work, which integrates machine learning, edge computing, and big data. Presents the basics of training machine learning models, key challenges and issues, as well as comprehensive techniques including edge learning algorithms, and system design issues. Describes architectures, frameworks, and key technologies for learning performance, security, and privacy, as well as incentive issues in training/inference at the network edge. Intended to stimulate fruitful discussions, inspire further research ideas, and inform readers from both academia and industry backgrounds. Essential reading for experienced researchers and developers, or for those who are just entering the field.

This edited volume on machine learning and big data analytics (Proceedings of ICMLBDA 2021) is intended to be used as a reference book for researchers and practitioners in the disciplines of computer science, electronics and telecommunication, information science, and electrical engineering. Machine learning and Big data analytics represent a key ingredients in the industrial applications for new products and services. Big data analytics applies machine learning for predictions by examining large and varied data sets-i.e., big data-to uncover hidden patterns, unknown correlations, market trends, customer preferences, and other useful information that can help organizations make more informed business decisions.

This collection of articles is devoted to fuzzy as well as rough set theories. Both theoriesarebasedonrigorousideas, methodsandtechniquesinlogic, mathem- ics, and computer science for treating problems for which approximate solutions are possible only, due to their inherent ambiguity, vagueness, incompleteness, etc. Vast areas of decision making, data mining, knowledge discovery in data, approximatereasoning, etc., aresuccessfully exploredusing methods workedout within fuzzy and rough paradigms. By the very nature of fuzzy and rough paradigms, outlined above, they are related to distinct logical schemes: it is well-known that rough sets are related to modal logicsS5andS4(Orl owska, E., Modal logics in the theory of infor- tion systems, Z. Math. Logik Grund. Math. 30, 1984, pp. 213 ?.; Vakarelov, D., Modal logics for knowledgerepresentationsystems, LNCS 363,1989, pp. 257?.) and to ?nitely-valued logics (Pagliani, P., Rough set theory and logic-algebraic structures. In Incomplete Information: Rough Set Analysis, Orlo wska, E., ed., Physica/Springer, 1998, pp. 109 ?.; Polkowski, L. A note on 3-valued rough logic accepting decision rules, Fundamenta Informaticae 61, to appear). Fuzzy sets are related to in?nitely-valued logics (fuzzy membership to degree r? 0,1]expressingtruthdegreer)(Goguen, J.A., Thelogicofinexactconcepts, Synthese18/19,1968-9, pp.325?.;Pavelka, J., OnfuzzylogicI, II, III, Z. Math. Logik Grund. Math. 25, 1979, pp. 45 ?., pp. 119 ?., pp. 454 ?.; Dubois, D., Prade, H., Possibility Theory, Plenum Press, 1988; Haj ek, P., Metamathematics of Fuzzy Logic, Kluw

The work described in this book was first presented at the Second Workshop on Genetic Programming, Theory and Practice, organized by the Center for the Study of Complex Systems at the University of Michigan, Ann Arbor, 13-15 May 2004. The goal of this workshop series is to promote the exchange of research results and ideas between those who focus on Genetic Programming (GP) theory and those who focus on the application of GP to various re- world problems. In order to facilitate these interactions, the number of talks and participants was small and the time for discussion was large. Further, participants were asked to review each other's chapters before the workshop. Those reviewer comments, as well as discussion at the workshop, are reflected in the chapters presented in this book. Additional information about the workshop, addendums to chapters, and a site for continuing discussions by participants and by others can be found at http: //cscs.umich.edu:8000/GPTP-20041. We thank all the workshop participants for making the workshop an exciting and productive three days. In particular we thank all the authors, without whose hard work and creative talents, neither the workshop nor the book would be possible. We also thank our keynote speakers Lawrence ("Dave") Davis of NuTech Solutions, Inc., Jordan Pollack of Brandeis University, and Richard Lenski of Michigan State University, who delivered three thought-provoking speeches that inspired a great deal of discussion among the participants.

Data mining (DM) consists of extracting interesting knowledge from re- world, large & complex data sets; and is the core step of a broader process, called the knowledge discovery from databases (KDD) process. In addition to the DM step, which actually extracts knowledge from data, the KDD process includes several preprocessing (or data preparation) and post-processing (or knowledge refinement) steps. The goal of data preprocessing methods is to transform the data to facilitate the application of a (or several) given DM algorithm(s), whereas the goal of knowledge refinement methods is to validate and refine discovered knowledge. Ideally, discovered knowledge should be not only accurate, but also comprehensible and interesting to the user. The total process is highly computation intensive. The idea of automatically discovering knowledge from databases is a very attractive and challenging task, both for academia and for industry. Hence, there has been a growing interest in data mining in several AI-related areas, including evolutionary algorithms (EAs). The main motivation for applying EAs to KDD tasks is that they are robust and adaptive search methods, which perform a global search in the space of candidate solutions (for instance, rules or another form of knowledge representation).

Algorithmic learning theory is mathematics about computer programs which learn from experience. This involves considerable interaction between various mathematical disciplines including theory of computation, statistics, and c- binatorics. There is also considerable interaction with the practical, empirical ?elds of machine and statistical learning in which a principal aim is to predict, from past data about phenomena, useful features of future data from the same phenomena. The papers in this volume cover a broad range of topics of current research in the ?eld of algorithmic learning theory. We have divided the 29 technical, contributed papers in this volume into eight categories (corresponding to eight sessions) re?ecting this broad range. The categories featured are Inductive Inf- ence, Approximate Optimization Algorithms, Online Sequence Prediction, S- tistical Analysis of Unlabeled Data, PAC Learning & Boosting, Statistical - pervisedLearning, LogicBasedLearning, andQuery&ReinforcementLearning. Below we give a brief overview of the ?eld, placing each of these topics in the general context of the ?eld. Formal models of automated learning re?ect various facets of the wide range of activities that can be viewed as learning. A ?rst dichotomy is between viewing learning as an inde?nite process and viewing it as a ?nite activity with a de?ned termination. Inductive Inference models focus on inde?nite learning processes, requiring only eventual success of the learner to converge to a satisfactory conclusion.

The proceedings of ECML/PKDD 2004 are published in two separate, albeit - tertwined, volumes: theProceedingsofthe 15thEuropeanConferenceonMac- ne Learning (LNAI 3201) and the Proceedings of the 8th European Conferences on Principles and Practice of Knowledge Discovery in Databases (LNAI 3202). The two conferences were co-located in Pisa, Tuscany, Italy during September 20-24, 2004. It was the fourth time in a row that ECML and PKDD were co-located. - ter the successful co-locations in Freiburg (2001), Helsinki (2002), and Cavtat- Dubrovnik (2003), it became clear that researchersstrongly supported the or- nization of a major scienti?c event about machine learning and data mining in Europe. We are happy to provide some statistics about the conferences. 581 di?erent papers were submitted to ECML/PKDD (about a 75% increase over 2003); 280 weresubmittedtoECML2004only,194weresubmittedtoPKDD2004only, and 107weresubmitted to both.Aroundhalfofthe authorsforsubmitted papersare from outside Europe, which is a clear indicator of the increasing attractiveness of ECML/PKDD. The Program Committee members were deeply involved in what turned out to be a highly competitive selection process. We assigned each paper to 3 - viewers, deciding on the appropriate PC for papers submitted to both ECML and PKDD. As a result, ECML PC members reviewed 312 papers and PKDD PC members reviewed 269 papers. We accepted for publication regular papers (45 for ECML 2004 and 39 for PKDD 2004) and short papers that were as- ciated with poster presentations (6 for ECML 2004 and 9 for PKDD 2004). The globalacceptance ratewas14.5%for regular papers(17% if we include the short paper

We would like to present, with great pleasure, the ?rst volume of a new jo- nal, Transactions on Rough Sets. This journal, part of the new journal subline in the Springer-Verlag series Lecture Notes in Computer Science, is devoted to the entire spectrum of rough set related issues, starting from logical and ma- ematical foundations of rough sets, through all aspects of rough set theory and its applications, data mining, knowledge discovery and intelligent information processing, to relations between rough sets and other approaches to uncertainty, vagueness, and incompleteness, such as fuzzy sets, theory of evidence, etc. The ?rst, pioneering papers on rough sets, written by the originator of the idea, ProfessorZdzis lawPawlak, werepublishedintheearly1980s.Weareproud to dedicate this volume to our mentor, Professor Zdzis law Pawlak, who kindly enriched this volume with his contribution on philosophical, logical, and mat- matical foundations of roughset theory. In his paper Professor Pawlakshows all over again the underlying ideas of rough set theory as well as its relations with Bayes' theorem, con?ict analysis, ?ow graphs, decision networks, and decision rules.

This volume contains papers presented at the 17th Annual Conference on Le- ning Theory (previously known as the Conference on Computational Learning Theory) held in Ban?, Canada from July 1 to 4, 2004. The technical program contained 43 papers selected from 107 submissions, 3 open problems selected from among 6 contributed, and 3 invited lectures. The invited lectures were given by Michael Kearns on 'Game Theory, Automated Trading and Social Networks', Moses Charikar on 'Algorithmic Aspects of - nite Metric Spaces', and Stephen Boyd on 'Convex Optimization, Semide?nite Programming, and Recent Applications'. These papers were not included in this volume. The Mark Fulk Award is presented annually for the best paper co-authored by a student. Thisyear theMark Fulk award wassupplemented with two further awards funded by the Machine Learning Journal and the National Information Communication Technology Centre, Australia (NICTA). We were therefore able toselectthreestudentpapersforprizes.ThestudentsselectedwereMagalieF- montforthesingle-authorpaper"ModelSelectionbyBootstrapPenalizationfor Classi?cation", Daniel Reidenbach for the single-author paper "On the Lear- bility of E-Pattern Languages over Small Alphabets", and Ran Gilad-Bachrach for the paper "Bayes and Tukey Meet at the Center Point" (co-authored with Amir Navot and Naftali Tishby).

Automatic Quantum Computer Programming provides an introduction to quantum computing for non-physicists, as well as an introduction to genetic programming for non-computer-scientists. The book explores several ways in which genetic programming can support automatic quantum computer programming and presents detailed descriptions of specific techniques, along with several examples of their human-competitive performance on specific problems. Source code for the author 's QGAME quantum computer simulator is included as an appendix, and pointers to additional online resources furnish the reader with an array of tools for automatic quantum computer programming.

The fusion of di?erent information sourcesis a persistent and intriguing issue. It hasbeenaddressedforcenturiesinvariousdisciplines, includingpoliticalscience, probability and statistics, system reliability assessment, computer science, and distributed detection in communications. Early seminal work on fusion was c- ried out by pioneers such as Laplace and von Neumann. More recently, research activities in information fusion have focused on pattern recognition. During the 1990s, classi?erfusionschemes, especiallyattheso-calleddecision-level, emerged under a plethora of di?erent names in various scienti?c communities, including machine learning, neural networks, pattern recognition, and statistics. The d- ferent nomenclatures introduced by these communities re?ected their di?erent perspectives and cultural backgrounds as well as the absence of common forums and the poor dissemination of the most important results. In 1999, the ?rst workshop on multiple classi?er systems was organized with the main goal of creating a common international forum to promote the diss- ination of the results achieved in the diverse communities and the adoption of a common terminology, thus giving the di?erent perspectives and cultural ba- grounds some concrete added value. After ?ve meetings of this workshop, there is strong evidence that signi?cant steps have been made towards this goal. - searchers from these diverse communities successfully participated in the wo- shops, and world experts presented surveys of the state of the art from the perspectives of their communities to aid cross-fertilizat

The field called Learning Classifier Systems is populated with romantics. Why shouldn't it be possible for computer programs to adapt, learn, and develop while interacting with their environments? In particular, why not systems that, like organic populations, contain competing, perhaps cooperating, entities evolving together? John Holland was one of the earliest scientists with this vision, at a time when so-called artificial intelligence was in its infancy and mainly concerned with preprogrammed systems that didn't learn. that, like organisms, had sensors, took Instead, Holland envisaged systems actions, and had rich self-generated internal structure and processing. In so doing he foresaw and his work prefigured such present day domains as reinforcement learning and embedded agents that are now displacing the older "standard Af' . One focus was what Holland called "classifier systems" sets of competing rule like "classifiers," each a hypothesis as to how best to react to some aspect of the environment--or to another rule. The system embracing such a rule "popu lation" would explore its available actions and responses, rewarding and rating the active rules accordingly. Then "good" classifiers would be selected and re produced, mutated and even crossed, a la Darwin and genetics, steadily and reliably increasing the system's ability to cope."

This four-volume set of LNCS 12821, LNCS 12822, LNCS 12823 and LNCS 12824, constitutes the refereed proceedings of the 16th International Conference on Document Analysis and Recognition, ICDAR 2021, held in Lausanne, Switzerland in September 2021. The 182 full papers were carefully reviewed and selected from 340 submissions, and are presented with 13 competition reports. The papers are organized into the following topical sections: scene text detection and recognition, document classification, gold-standard benchmarks and data sets, historical document analysis, and handwriting recognition. In addition, the volume contains results of 13 scientific competitions held during ICDAR 2021.

This four-volume set of LNCS 12821, LNCS 12822, LNCS 12823 and LNCS 12824, constitutes the refereed proceedings of the 16th International Conference on Document Analysis and Recognition, ICDAR 2021, held in Lausanne, Switzerland in September 2021. The 182 full papers were carefully reviewed and selected from 340 submissions, and are presented with 13 competition reports. The papers are organized into the following topical sections: extracting document semantics, text and symbol recognition, document analysis systems, office automation, signature verification, document forensics and provenance analysis, pen-based document analysis, human document interaction, document synthesis, and graphs recognition.