Studienarbeit aus dem Jahr 2020 im Fachbereich Informatik - Künstliche Intelligenz, Note: 1.0, FernUniversität Hagen (Lehrstuhl für Parallelverarbeitung und IT-Sicherheit), Sprache: Deutsch, Abstract: Die Defense Advanced Research Projects Agency (DARPA) des Verteidigungsministeriums der Vereinigten Staaten von Amerika rief 2016 das Explainable Artificial Intelligence (XAI)-Programm ins Leben, mit dem Fokus, Techniken maschinellen Lernens zu entwickeln, die erstens erklärbare Modelle bei gleichbleibend hoher Lernfähigkeit erzeugen und zweitens den Menschen befähigen, AI-Systeme zu verstehen, ihnen angemessen zu vertrauen und die nächste Generation intelligenter Systeme kontrollieren zu können. Denn nur, wenn wir Menschen verstehen, wie KI-Systeme zu Entscheidungen gelangen, haben wir die Möglichkeit, sie mehrwertbringend in industriellen Prozessen anzuwenden und positiv in das tägliche Leben unserer Gesellschaft zu integrieren. Zielvorgabe der vorliegenden Arbeit ist die Darstellung und Analyse zweier Methoden, namentlich der Sensitivity Analysis (SA) und der Layerwise Relevance Propagation (LRP), deren eigene Zielsetzung es ist, die Entscheidungen intelligenter Systeme für den menschlichen Betrachter nachvollziehbar zu machen. Bevor jedoch auf die einzelnen Methoden, deren Funktionsweise und auf eine kritischen Betrachtung eingegangen werden kann, bedarf es einer Abgrenzung des Forschungsgebiets der XAI. Artificial Intelligence (AI) – ein populärer Begriff aktueller Zeit, der mit einer Vielzahl an technologischen Anwendungen in Verbindung gebracht wird. Manchmal prominent und tangibel in Form von menschenähnlichen Robotern, wieder andere Male subtil, weder sichtbar noch greifbar in Form von Algorithmen. Dabei ist die konzeptionelle Idee der Abbildung künstlicher Intelligenz durch Computersysteme keine neue Errungenschaft, sondern geht bis auf die Erkenntnisse von Warren McCulloch und Walter Pitts im Jahr 1943 zurück. McCulloch/Pitts (1943) offerierten ein Modell, das in Anlehnung an das biologische Vorbild künstliche Neuronen an- bzw. ausschaltet, je nach Stimulus durch benachbarte Neuronen. Dabei wird impliziert, dass durch ein Netzwerk künstlicher Neuronen ebenfalls die Möglichkeit der Lernfähigkeit bestünde.

This book is about how computer systems might be designed to serve their users rather better. It deals with how to study the natural behaviour of users to see how computer systems might best help them, and how one might also involve them in the design of computer systems that will assist them in their everyday practices.

'Future-Proofing' addresses the problems of sustainability in IT research projects. It provides a conceptual framework which allows readers to better understand sustainability issues, make them aware of the challenges around effective sustainability, and provide tangible suggestions for researchers to put into action.

The field of knowledge management focuses on how organizations can most effectively store, manage, retrieve, and enlarge their intellectual properties. The repository view of knowledge management emphasizes the gathering, providing, and filtering of explicit knowledge. The information in a repository has the advantage of being easily transferable and reusable. But it is not easy to use decontextualized information, and users often need access to human experts. This book describes a more recent approach to knowledge management, which the authors call "expertise sharing." Expertise sharing emphasizes the human aspects -- cognitive, social, cultural, and organizational -- of knowledge management, in addition to information storage and retrieval. Rather than focusing on the management level of an organization, expertise sharing focuses on the self-organized activities of the organization's members. The book addresses the concerns of both researchers and practitioners, describing current literature and research as well as offering information on implementing systems. It consists of three parts: an introduction to knowledge sharing in large organizations; empirical studies of expertise sharing in different types of settings; and detailed descriptions of computer systems that can route queries, assemble people and work, and augment naturally occurring social networks within organizations.

A diverse group of scholars redefine constructionism--introduced by Seymour Papert in 1980--in light of new technologies and theories. Constructionism, first introduced by Seymour Papert in 1980, is a framework for learning to understand something by making an artifact for and with other people. A core goal of constructionists is to respect learners as creators, to enable them to engage in making meaning for themselves through construction, and to do this by democratizing access to the world's most creative and powerful tools. In this volume, an international and diverse group of scholars examine, reconstruct, and evolve the constructionist paradigm in light of new technologies and theories.

This book features state-of-the-art contributions from two well-established conferences: Changeable, Agile, Reconfigurable and Virtual Production Conference (CARV2020) and Mass Customization and Personalization Conference (MCPC2020). Together, they focus on the joint design, development, and management of products, production systems, and business for sustainable customization and personalization. The book covers a large range of topics within this domain, ranging from industrial success factors to original contributions within the field.

Made to Work analyses the conditions of mobile knowledge work (MKW) in contemporary worklives, contrasting and drawing parallels among three highly significant sectors of the Knowledge Economy: academia, information communication technology (ICT) management, and digital creative work. It introduces the concept of ‘corollary work’ to characterise the elusive work underpinning the configuration of workers, informational, technological, relational and infrastructural resources in (re)producing liveable worklives. It ultimately illuminates the myriad strands of corollary work that enable MKW to take place and contributes to emergent debates on how exploitation, at least in the domain of MKW, can be named, resisted and creatively subverted. In so doing, it opens up a conversation about the complex ways in which contemporary worklives are ‘made to work’, and about potential interventions to bring about more just worklife conditions in the future.