SpringerBriefs in Computer Science – serie

The ultimate goal of machines is to help humans to solve problems.Such problems range between two extremes: structured problems for which the solution is totally defined (and thus are easily programmed by humans), and random problems for which the solution is completely undefined (and thus cannot be programmed). Problems in the vast middle ground have solutions that cannot be well defined and are, thus, inherently hard to program. Machine Learning is the way to handle this vast middle ground, so that many tedious and difficult hand-coding tasks would be replaced by automatic learning methods. There are several machine learning tasks, and this work is focused on a major one, which is known as classification. Some classification problems are hard to solve, but we show that they can be decomposed into much simpler sub-problems. We also show that independently solving these sub-problems by taking into account their particular demands, often leads to improved classification performance.

Voice over IP (VoIP) and Internet Multimedia Subsystem technologies (IMS) are rapidly being adopted by consumers, enterprises, governments and militaries. In this book, the authors examine the current state of affairs on VoIP security through a survey of 221 known/disclosed security vulnerabilities in bug-tracking databases.

As computer systems evolve, the volume of data to be processed increases significantly, either as a consequence of the expanding amount of available information, or due to the possibility of performing highly complex operations that were not feasible in the past. Nevertheless, tasks that depend on the manipulation of large amounts of information are still performed at large computational cost, i.e., either the processing time will be large, or they will require intensive use of computer resources. In this scenario, the efficient use of available computational resources is paramount, and creates a demand for systems that can optimize the use of resources in relation to the amount of data to be processed. This problem becomes increasingly critical when the volume of information to be processed is variable, i.e., there is a seasonal variation of demand. Such demand variations are caused by a variety of factors, such as an unanticipated burst of client requests, a time-critical simulation,or high volumes of simultaneous video uploads, e.g. as a consequence of a public contest. In these cases, there are moments when the demand is very low (resources are almost idle) while, conversely, at other moments, the processing demand exceeds the resources capacity. Moreover, from an economical perspective, seasonal demands do not justify a massive investment in infrastructure, just to provide enough computing power for peak situations. In this light, the ability to build adaptive systems, capable of using on demand resources provided by Cloud Computing infrastructures is very attractive.

The book describes the world's first successful experiment in fully automated board game design. Evolutionary methods were used to derive new rule sets within a custom game description language, and self-play trials used to estimate each derived game's potential to interest human players. The end result is a number of new and interesting games, one of which has proved popular and gone on to be commercially published.

Since their first inception, automatic reading systems have evolved substantially, yet the recognition of handwriting remains an open research problem due to its substantial variation in appearance. With the introduction of Markovian models to the field, a promising modeling and recognition paradigm was established for automatic handwriting recognition. However, no standard procedures for building Markov model-based recognizers have yet been established. This text provides a comprehensive overview of the application of Markov models in the field of handwriting recognition, covering both hidden Markov models and Markov-chain or n-gram models. First, the text introduces the typical architecture of a Markov model-based handwriting recognition system, and familiarizes the reader with the essential theoretical concepts behind Markovian models. Then, the text reviews proposed solutions in the literature for open problems in applying Markov model-based approaches to automatic handwriting recognition.

This book is a guide to getting started with ILDJIT, a compilation framework designed to be both easily extensible and easily configurable. Within this framework, it is possible to build a tool-chain by customizing ILDJIT for specific purposes. Customizations can be used within both static and dynamic compilers already included in the framework without adaptations. Moreover, customizations allow modification of both the behaviors and the characteristics of these compilers to better satisfy the particular need. Currently, ILDJIT is able to translate bytecode programs to generate machine code for both Intel x86 and ARM processors. By relying on ILDJIT technology, more input languages or platforms can be supported. After an introduction to ILDJIT, this guide goes into detail on how to exploit it by extending the framework to match specific requirements. Finally, there is an introduction and discussion of the design choices followed during the authors’ years of development efforts towards ILDJIT.

Image compression is concerned with minimization of the number of information carrying units used to represent an image. Lossy compression techniques incur some loss of information which is usually imperceptible. In return for accepting this distortion, we obtain much higher compression ratios than is possible with lossless compression. Salient features of this book include: four new image compression algorithms and implementation of these algorithms; detailed discussion of fuzzy geometry measures and their application in image compression algorithms; new domain decomposition based algorithms using image quality measures and study of various quality measures for gray scale image compression; compression algorithms for different parallel architectures and evaluation of time complexity for encoding on all architectures; parallel implementation of image compression algorithms on a cluster in Parallel Virtual Machine (PVM) environment.

This work presents a full generic approach to the detection and recognition of traffic signs. The approach is based on the latest computer vision methods for object detection, and on powerful methods for multiclass classification. The challenge was to robustly detect a set of different sign classes in real time, and to classify each detected sign into a large, extensible set of classes. To address this challenge, several state-of-the-art methods were developed that can be used for different recognition problems. Following an introduction to the problems of traffic sign detection and categorization, the text focuses on the problem of detection, and presents recent developments in this field. The text then surveys a specific methodology for the problem of traffic sign categorization – Error-Correcting Output Codes – and presents several algorithms, performing experimental validation on a mobile mapping application. The work ends with a discussion on future research and continuing challenges.

Service provisioning in ad hoc networks is challenging given the difficulties of communicating over a wireless channel and the potential heterogeneity and mobility of the devices that form the network. Service placement is the process of selecting an optimal set of nodes to host the implementation of a service in light of a given service demand and network topology. The key advantage of active service placement in ad hoc networks is that it allows for the service configuration to be adapted continuously at run time. Service Placement in Ad Hoc Networks proposes the SPi service placement framework as a novel approach to service placement in ad hoc networks. The SPi framework takes advantage of the interdependencies between service placement, service discovery and the routing of service requests to minimize signaling overhead. The work also proposes the Graph Cost / Single Instance and the Graph Cost / Multiple Instances placement algorithms.

Software similarity and classification is an emerging topic with wide applications. It is applicable to the areas of malware detection, software theft detection, plagiarism detection, and software clone detection. Extracting program features, processing those features into suitable representations, and constructing distance metrics to define similarity and dissimilarity are the key methods to identify software variants, clones, derivatives, and classes of software. Software Similarity and Classification reviews the literature of those core concepts, in addition to relevant literature in each application and demonstrates that considering these applied problems as a similarity and classification problem enables techniques to be shared between areas. Additionally, the authors present in-depth case studies using the software similarity and classification techniques developed throughout the book.

Entropy Guided Transformation Learning: Algorithms and Applications (ETL) presents a machine learning algorithm for classification tasks. ETL generalizes Transformation Based Learning (TBL) by solving the TBL bottleneck: the construction of good template sets. ETL automatically generates templates using Decision Tree decomposition.The authors describe ETL Committee, an ensemble method that uses ETL as the base learner. Experimental results show that ETL Committee improves the effectiveness of ETL classifiers. The application of ETL is presented to four Natural Language Processing (NLP) tasks: part-of-speech tagging, phrase chunking, named entity recognition and semantic role labeling. Extensive experimental results demonstrate that ETL is an effective way to learn accurate transformation rules, and shows better results than TBL with handcrafted templates for the four tasks. By avoiding the use of handcrafted templates, ETL enables the use of transformation rules to a greater range of tasks.Suitable for both advanced undergraduate and graduate courses, Entropy Guided Transformation Learning: Algorithms and Applications provides a comprehensive introduction to ETL and its NLP applications.

Description Logics (DLs) is a family of formalisms used to represent knowledge of a domain. They are equipped with a formal logic-based semantics. Knowledge representation systems based on description logics provide various inference capabilities that deduce implicit knowledge from the explicitly represented knowledge.A Proof Theory for Description Logics introduces Sequent Calculi and Natural Deduction for some DLs (ALC, ALCQ). Cut-elimination and Normalization are proved for the calculi. The author argues that such systems can improve the extraction of computational content from DLs proofs for explanation purposes.

Research in context-aware computing has produced a number of middleware systems for context management. However, development of ubiquitous context-aware applications is still a challenge because most current middleware systems are still focused on isolated and static context-aware environments. Context-aware environments are inherently dynamic as a result of occasional additions or upgrade of sensors, applications or context inference mechanisms.Context Management for Distributed and Dynamic Context-Aware Computing proposes a novel architecture for context management based on the concept of context domains, allowing applications to keep context interests across distributed context management systems. The authors describe a distributed middleware that implements the aforementioned concepts, without compromising scalability and efficiency of context access.

In Ambient Intelligence (AmI) systems, reasoning is fundamental for triggering actions or adaptations according to specific situations that may be meaningful and relevant to some applications. However, such reasoning operations may need to evaluate context data collected from distributed sources and stored in different devices, as usually not all context data is readily available to the reasoners within the system. Decentralized Reasoning in Ambient Intelligence proposes a decentralized reasoning approach for performing rule-based reasoning about context data targeting AmI systems. For this purpose, the authors define a context model assuming context data distributed over two sides: the user side, represented by the users and their mobile devices, and the ambient side, represented by the fixed computational infrastructure and ambient services. They formalize the cooperative reasoning operation — in which two entities cooperate to perform decentralized rule-based reasoning — and define a complete process to perform this operation.

however, classical belief revision is not well suited for logics such as Description Logics.Belief Revision in Non-Classical Logics presents a framework which can be applied to a wide class of logics that include – besides most Description Logics such as the ones behind OWL – Horn Logic and Intuitionistic logic, amongst others.

The network management community has been pushed towards the design of alternative management approaches able to support heterogeneity, scalability, reliability, and minor human intervention. The employment of self-* properties and Peer-To-Peer (P2P) are seen as promising alternatives, able to provide the sophisticated solutions required. Despite being developed in parallel, and with minor direct connections perceived between them, self-* properties and P2P can be used concurrently. In Self-* and P2P for Network Management: Design Principles and Case Studies, the authors explore the issues behind the joint use of self-* properties and P2P, and present: a survey relating autonomic computing and self-* properties, P2P, and network and service management; the design of solutions that explore parallel and cooperative behavior of management peers; the change in angle of network management solution development from APIs, protocols, architectures, and frameworks to the design of management algorithms.

This work proposes a complete sensor-independent visual system that provides robust target motion detection. First, the way sensors obtain images, in terms of resolution distribution and pixel neighbourhood, is studied. This allows a spatial analysis of motion to be carried out. Then, a novel background maintenance approach for robust target motion detection is implemented. Two different situations are considered: a fixed camera observing a constant background where objects are moving; and a still camera observing objects in movement within a dynamic background. This distinction lies on developing a surveillance mechanism without the constraint of observing a scene free of foreground elements for several seconds when a reliable initial background model is obtained, as that situation cannot be guaranteed when a robotic system works in an unknown environment. Other problems are also addressed to successfully deal with changes in illumination, and the distinction between foreground and background elements.

Change detection using remotely sensed images has many applications, such as urban monitoring, land-cover change analysis, and disaster management. This work investigates two-dimensional change detection methods. The existing methods in the literature are grouped into four categories: pixel-based, transformation-based, texture analysis-based, and structure-based. In addition to testing existing methods, four new change detection methods are introduced: fuzzy logic-based, shadow detection-based, local feature-based, and bipartite graph matching-based. The latter two methods form the basis for a structural analysis of change detection. Three thresholding algorithms are compared, and their effects on the performance of change detection methods are measured. These tests on existing and novel change detection methods make use of a total of 35 panchromatic and multi-spectral Ikonos image sets. Quantitative test results and their interpretations are provided.

This text provides a survey of the latest research into the Hough transform for line detection and its contemporary developments. Written with students in mind, this work can serve as a condensed textbook and as a reference for practitioners working in computer vision. The text also presents an overview and detailed elaboration of recent research related to PClines – a line parameterization based on parallel coordinates. A detailed analysis of the concept is given, including implementation details and measurements. One chapter discusses the detection of chessboard-like patterns, and sets of parallel lines and lines coincident with one (vanishing) point – a topic with many applications, such as 2D barcodes, aerial images, and architecture recognition. The work summarizes recent research in the field, and analyses new advances achieved using the PClines parameterization.

In this work, the unique power measurement capabilities of the Cray XT architecture were exploited to gain an understanding of power and energy use, and the effects of tuning both CPU and network bandwidth. Modifications were made to deterministically halt cores when idle. Additionally, capabilities were added to alter operating P-state. At the application level, an understanding of the power requirements of a range of important DOE/NNSA production scientific computing applications running at large scale is gained by simultaneously collecting current and voltage measurements on the hosting nodes. The effects of both CPU and network bandwidth tuning are examined, and energy savings opportunities without impact on run-time performance are demonstrated. This research suggests that next-generation large-scale platforms should not only approach CPU frequency scaling differently, but could also benefit from the capability to tune other platform components to achieve more energy-efficient performance.

This user’s reference is a companion to the separate book also titled “Guide to Modelling and Simulation of Systems of Systems.” The principal book explicates integrated development environments to support virtual building and testing of systems of systems, covering in some depth the MS4 Modelling Environment™. This user’s reference provides a quick reference and exposition of the various concepts and functional features covered in that book. The topics in the user’s reference are grouped in alignment with the workflow displayed on the MS4 Modeling Environment™ launch page, under the headings Atomic Models, System Entity Structure, Pruning SES, and Miscellaneous. For each feature, the reference discusses why we use it, when we should use it, and how to use it. Further comments and links to related features are also included.

Time-of-flight (TOF) cameras provide a depth value at each pixel, from which the 3D structure of the scene can be estimated. This new type of active sensor makes it possible to go beyond traditional 2D image processing, directly to depth-based and 3D scene processing. Many computer vision and graphics applications can benefit from TOF data, including 3D reconstruction, activity and gesture recognition, motion capture and face detection. It is already possible to use multiple TOF cameras, in order to increase the scene coverage, and to combine the depth data with images from several colour cameras. Mixed TOF and colour systems can be used for computational photography, including full 3D scene modelling, as well as for illumination and depth-of-field manipulations. This work is a technical introduction to TOF sensors, from architectural and design issues, to selected image processing and computer vision methods.

Human action analysis and recognition is a relatively mature field, yet one which is often not well understood by students and researchers.  The large number of possible variations in human motion and appearance, camera viewpoint, and environment, present considerable challenges.  Some important and common problems remain unsolved by the computer vision community. However, many valuable approaches have been proposed over the past decade, including the motion history image (MHI) method. This method has received significant attention, as it offers greater robustness and performance than other techniques. This work presents a comprehensive review of these state-of-the-art approaches and their applications, with a particular focus on the MHI method and its variants.

The amount and the complexity of the data gathered by current enterprises are increasing at an exponential rate. Consequently, the analysis of Big Data is nowadays a central challenge in Computer Science, especially for complex data. For example, given a satellite image database containing tens of Terabytes, how can we find regions aiming at identifying native rainforests, deforestation or reforestation? Can it be made automatically? Based on the work discussed in this book, the answers to both questions are a sound “yes”, and the results can be obtained in just minutes. In fact, results that used to require days or weeks of hard work from human specialists can now be obtained in minutes with high precision. Data Mining in Large Sets of Complex Data discusses new algorithms that take steps forward from traditional data mining (especially for clustering) by considering large, complex datasets. Usually, other works focus in one aspect, either data size or complexity. This work considers both: it enables mining complex data from high impact applications, such as breast cancer diagnosis, region classification in satellite images, assistance to climate change forecast, recommendation systems for the Web and social networks; the data are large in the Terabyte-scale, not in Giga as usual; and very accurate results are found in just minutes. Thus, it provides a crucial and well timed contribution for allowing the creation of real time applications that deal with Big Data of high complexity in which mining on the fly can make an immeasurable difference, such as supporting cancer diagnosis or detecting deforestation.

This work presents a study of cache replacement strategies designed for static web content. Proxy servers can improve performance by caching static web content such as cascading style sheets, java script source files, and large files such as images. This topic is particularly important in wireless ad hoc networks, in which mobile devices act as proxy servers for a group of other mobile devices. Opening chapters present an introduction to web requests and the characteristics of web objects, web proxy servers and Squid, and artificial neural networks. This is followed by a comprehensive review of cache replacement strategies simulated against different performance metrics. The work then describes a novel approach to web proxy cache replacement that uses neural networks for decision making, evaluates its performance and decision structures, and examines its implementation in a real environment, namely, in the Squid proxy server.

Due to its inherent time-scale locality characteristics, the discrete wavelet transform (DWT) has received considerable attention in signal/image processing. Wavelet transforms have excellent energy compaction characteristics and can provide perfect reconstruction. The shifting (translation) and scaling (dilation) are unique to wavelets. Orthogonality of wavelets with respect to dilations leads to multigrid representation. As the computation of DWT involves filtering, an efficient filtering process is essential in DWT hardware implementation. In the multistage DWT, coefficients are calculated recursively, and in addition to the wavelet decomposition stage, extra space is required to store the intermediate coefficients. Hence, the overall performance depends significantly on the precision of the intermediate DWT coefficients. This work presents new implementation techniques of DWT, that are efficient in terms of computation, storage, and with better signal-to-noise ratio in the reconstructed signal.

This work focuses on central catadioptric systems, from the early step of calibration to high-level tasks such as 3D information retrieval. The book opens with a thorough introduction to the sphere camera model, along with an analysis of the relation between this model and actual central catadioptric systems. Then, a new approach to calibrate any single-viewpoint catadioptric camera is described.  This is followed by an analysis of existing methods for calibrating central omnivision systems, and a detailed examination of hybrid two-view relations that combine images acquired with uncalibrated central catadioptric systems and conventional cameras. In the remaining chapters, the book discusses a new method to compute the scale space of any omnidirectional image acquired with a central catadioptric system, and a technique for computing the orientation of a hand-held omnidirectional catadioptric camera.

This book introduces the statistical software R to the image processing community in an intuitive and practical manner. R brings interesting statistical and graphical tools which are important and necessary for image processing techniques. Furthermore, it has been proved in the literature that R is among the most reliable, accurate and portable statistical software available. Both the theory and practice of R code concepts and techniques are presented and explained, and the reader is encouraged to try their own implementation to develop faster, optimized programs. Those who are new to the field of image processing and to R software will find this work a useful introduction. By reading the book alongside an active R session, the reader will experience an exciting journey of learning and programming.

There is a significant difference between designing a new algorithm, proving its correctness, and teaching it to an audience. This work provides a set of novel metaphors identified and developed as suitable tools for teaching many of the "classic textbook" algorithms taught in undergraduate courses worldwide.