„Forscher warnen: Künstliche Intelligenz wird wahrscheinlich die Menschheit auslöschen“ (Schmidt 2022) – solche Headlines begegnen uns im Diskurs über künstliche Intelligenz immer wieder. Reißerische Titel, die die Übermacht der Maschinen und den damit einhergehenden Untergang des Menschen postulieren. Sie schüren Angst und Misstrauen gegenüber neuen Technologien – und verhindern somit gegebenenfalls die Bereitschaft zu Fortschritt und Wandel. Entwicklungen im Bereich der Künstlichen Intelligenz werden in der Bevölkerung immer noch mit Misstrauen beobachtet. Das liegt vor allem daran, dass es sich bei diesem Forschungsfeld um ein komplexes handelt und in der öffentlichen Kommunikation häufig zur vermeintlichen Vereinfachung eine vermenschlichende Darstellungsform von Künstlicher Intelligenz gewählt wird. Es fehlen adäquate Beschreibungsformen und Sprachbilder, auf die nicht nur Medienvertreter:innen, sondern auch Wissenschaftler:innen zurückgreifen können. Womöglich unpassende vorhandene Metaphern und Sprachbilder, die die öffentliche Sicht auf das prägen, was sie beschreiben sollen, werden immer wieder bemüht. Diese unreflektierte Wiederholung führt zu einer starken Verankerung der Sprachmuster und damit zur Verankerung von irreführenden Konzeptualisierungen von Künstlicher Intelligenz. Derartige Sprachgebrauchsmuster formen den gesamtgesellschaftlichen Diskurs und leiten ihn somit gegebenenfalls in eine ungewollte Richtung. Neue Entwicklungen und Forschungsergebnisse kommunizieren die Wissenschaftler:innen häufig nicht direkt an die Öffentlichkeit, sondern oft sind Medien zwischengeschaltet. Der Informationsaustausch birgt somit (wie beim Stille-Post-Spielen) Risiken für Missverständnisse oder ungenaue Informationsweitergabe. Es ist auch Aufgabe der Wissenschaft, das eigene Kommunikationsverhalten zu reflektieren und die Öffentlichkeit über sie betreffende Entwicklungen auf angemessene Weise aufzuklären. Dafür bietet die Linguistik Unterstützung. Sie kann anderen Disziplinen Werkzeuge und Empfehlungen an die Hand geben, um ihren Wissensaustausch mit der Gesellschaft zu reflektieren und zu verbessern. Die Gründung des Tübinger Center for Rhetorical Science Communication Research on Artificial Intelligence (RHET AI) im Jahr 2021 zeigt, wie aktuell und wichtig diese Aufgabe ist. Immer mehr Linguist:innen beschäftigen sich mit Künstlicher Intelligenz, was auch auf die „Verwandtschaft“ zwischen den beiden Disziplinen zurückzuführen ist. Auch auf diese Beziehung wird im Folgenden eingegangen.

The capability of changing beliefs upon new information in a rational and efficient way is crucial for an intelligent agent. Belief change therefore is one of the central research fields in Artificial Intelligence (AI) for over two decades. In the AI literature, two different kinds of belief change operations have been intensively investigated: belief update, which deal with situations where the new information describes changes of the world; and belief revision, which assumes the world is static. As another important research area in AI, reasoning about actions mainly studies the problem of representing and reasoning about effects of actions. These two research fields are closely related and apply a common underlying principle, that is, an agent should change its beliefs (knowledge) as little as possible whenever an adjustment is necessary. This lays down the possibility of reusing the ideas and results of one field in the other, and vice verse. This thesis aims to develop a general framework and devise computational models that are applicable in reasoning about actions. Firstly, I shall propose a new framework for iterated belief revision by introducing a new postulate to the existing AGM/DP postulates, which provides general criteria for the design of iterated revision operators. Secondly, based on the new framework, a concrete iterated revision operator is devised. The semantic model of the operator gives nice intuitions and helps to show its satisfiability of desirable postulates. I also show that the computational model of the operator is almost optimal in time and space-complexity. In order to deal with the belief change problem in multi-agent systems, I introduce a concept of mutual belief revision which is concerned with information exchange among agents. A concrete mutual revision operator is devised by generalizing the iterated revision operator. Likewise, a semantic model is used to show the intuition and many nice properties of the mutual revision operator, and the complexity of its computational model is formally analyzed. Finally, I present a belief update operator, which takes into account two important problems of reasoning about action, i.e., disjunctive updates and domain constraints. Again, the updated operator is presented with both a semantic model and a computational model.

No description available.

The goal of General Game Playing (GGP) is to develop a system, that is able to automatically play previously unseen games well, solely by being given the rules of the game. In contrast to traditional game playing programs, a general game player cannot be given game specific knowledge. Instead, the program has to discover this knowledge and use it for effectively playing the game well without human intervention. In this thesis, we present a such a program and general methods that solve a variety of knowledge discovery problems in GGP. Our main contributions are methods for the automatic construction of heuristic evaluation functions, the automated discovery of game structures, a system for proving properties of games, and symmetry detection and exploitation for general games.

This thesis focuses on improving the SAT solving technology. The improvements focus on two major subjects: sequential SAT solving and parallel SAT solving. To better understand sequential SAT algorithms, the abstract reduction system Generic CDCL is introduced. With Generic CDCL, the soundness of solving techniques can be modeled. Next, the conflict driven clause learning algorithm is extended with the three techniques local look-ahead, local probing and all UIP learning that allow more global reasoning during search. These techniques improve the performance of the sequential SAT solver Riss. Then, the formula simplification techniques bounded variable addition, covered literal elimination and an advanced cardinality constraint extraction are introduced. By using these techniques, the reasoning of the overall SAT solving tool chain becomes stronger than plain resolution. When using these three techniques in the formula simplification tool Coprocessor before using Riss to solve a formula, the performance can be improved further. Due to the increasing number of cores in CPUs, the scalable parallel SAT solving approach iterative partitioning has been implemented in Pcasso for the multi-core architecture. Related work on parallel SAT solving has been studied to extract main ideas that can improve Pcasso. Besides parallel formula simplification with bounded variable elimination, the major extension is the extended clause sharing level based clause tagging, which builds the basis for conflict driven node killing. The latter allows to better identify unsatisfiable search space partitions. Another improvement is to combine scattering and look-ahead as a superior search space partitioning function. In combination with Coprocessor, the introduced extensions increase the performance of the parallel solver Pcasso. The implemented system turns out to be scalable for the multi-core architecture. Hence iterative partitioning is interesting for future parallel SAT solvers. The implemented solvers participated in international SAT competitions. In 2013 and 2014 Pcasso showed a good performance. Riss in combination with Copro- cessor won several first, second and third prices, including two Kurt-Gödel-Medals. Hence, the introduced algorithms improved modern SAT solving technology.

Mechanical ventilation can be a life saving treatment. However, due to the inhomogeneous and anisotropic behavior of the lung tissue, ventilation can also lead to overdistensions of lung regions whereas other areas remain even collapsed. A first step is a more comprehensive understanding of the flow mechanics under normal breathing conditions in a healthy lung as well as for a diseased, collapsed lung. This is the aim of this work. Therefore, a realistic model of the upper human airways has been generated at which experimental and numerical investigations could be carried out. Experimentally, the flow was analyzed by means of Particle Image Velocimetry (PIV) measurements which revealed new details about the flow patterns occurring during different ventilation frequencies. Numerical results were in good agreement with the experimental results and could provide new details about the three-dimensional flow structure and emerging secondary flow within the upper airways. The study of reopening of collapsed airways has shown that larger frequencies lead to airway reopening without overdistension of already open parts. Higher frequencies also lead to homogenization of mass flow distribution within the human lung. / Künstliche Beatmung ist meist eine lebensrettende Maßnahme. Aufgrund der räumlich anisotropen und inhomogenen Eigenschaften der Lunge kann die Beatmung jedoch auch zu einer Schädigung der Lunge führen. Daraus ergibt sich die Forderung einer „Protektiven Beatmung“. Ein erster Schritt dahingehend ist ein verbessertes Verständnis der Atmung und Beatmung am Beispiel der gesunden sowie kranken, teilweise kollabierten Lunge. Dies ist das Ziel der Arbeit. Hierfür wurde ein realistisches Modell der oberen Atemwege (Tracheobronchialbaum) angefertigt. An diesem Modell können sowohl experimentelle als auch numerische Untersuchungen durchgeführt werden. Experimentell wurde die Strömung mittels Particle Image Velocimetry (PIV) untersucht, wobei neue Details bezüglich der auftretenden Strömungsmuster für unterschiedliche Frequenzen gefunden wurden. Numerische Strömungsberechnungen stimmen gut mit den experimentellen Ergebnissen überein. Dreidimensionale Strömungsstrukturen sowie die Entwicklung von Sekundärwirbeln in der Lunge konnten erklärt werden. Eine Studie am kranken, teilweise kollabierten Lungenmodell zeigte, dass mit steigender Frequenz kollabierte Bereiche wiedereröffnet werden können. Höhere Frequenzen führen weiterhin zu einer Homogenisierung der Massenstromverteilung in der Lunge.

An autonomous robot that acts in a goal-directed fashion requires a world model of the elements that are relevant to the robot's task. In real-world, dynamic environments, the world model has to be created and continually updated from uncertain sensor data. The symbols used in plan-based robot control have to be anchored to detected objects. Furthermore, robot perception is not only a bottom-up and passive process: Knowledge about the composition of compound objects can be used to recognize larger-scale structures from their parts. Knowledge about the spatial context of an object and about common relations to other objects can be exploited to improve the quality of the world model and can inform an active search for objects that are missing from the world model. This thesis makes several contributions to address these challenges: First, a model-based semantic mapping system is presented that recognizes larger-scale structures like furniture based on semantic descriptions in an ontology. Second, a context-aware anchoring process is presented that creates and maintains the links between object symbols and the sensor data corresponding to those objects while exploiting the geometric context of objects. Third, an active perception system is presented that actively searches for a required object while being guided by the robot's knowledge about the environment.

This thesis addresses the question of how to teach dynamic motor skills to synthetic humanoids. A general approach based on imitation learning is presented and evaluated on a number of synthetic humanoids, as well as a number of different motor skills. The approach allows for intuitive and natural specification of motor skills without the need for expert knowledge. Using this approach we show that various important problems in robotics and computer animation can be tackled, including the synthesis of natural grasping, the synthesis of locomotion behavior or the physical interaction between humans and robots.

The capability of changing beliefs upon new information in a rational and efficient way is crucial for an intelligent agent. Belief change therefore is one of the central research fields in Artificial Intelligence (AI) for over two decades. In the AI literature, two different kinds of belief change operations have been intensively investigated: belief update, which deal with situations where the new information describes changes of the world; and belief revision, which assumes the world is static. As another important research area in AI, reasoning about actions mainly studies the problem of representing and reasoning about effects of actions. These two research fields are closely related and apply a common underlying principle, that is, an agent should change its beliefs (knowledge) as little as possible whenever an adjustment is necessary. This lays down the possibility of reusing the ideas and results of one field in the other, and vice verse. This thesis aims to develop a general framework and devise computational models that are applicable in reasoning about actions. Firstly, I shall propose a new framework for iterated belief revision by introducing a new postulate to the existing AGM/DP postulates, which provides general criteria for the design of iterated revision operators. Secondly, based on the new framework, a concrete iterated revision operator is devised. The semantic model of the operator gives nice intuitions and helps to show its satisfiability of desirable postulates. I also show that the computational model of the operator is almost optimal in time and space-complexity. In order to deal with the belief change problem in multi-agent systems, I introduce a concept of mutual belief revision which is concerned with information exchange among agents. A concrete mutual revision operator is devised by generalizing the iterated revision operator. Likewise, a semantic model is used to show the intuition and many nice properties of the mutual revision operator, and the complexity of its computational model is formally analyzed. Finally, I present a belief update operator, which takes into account two important problems of reasoning about action, i.e., disjunctive updates and domain constraints. Again, the updated operator is presented with both a semantic model and a computational model.

The goal of General Game Playing (GGP) is to develop a system, that is able to automatically play previously unseen games well, solely by being given the rules of the game. In contrast to traditional game playing programs, a general game player cannot be given game specific knowledge. Instead, the program has to discover this knowledge and use it for effectively playing the game well without human intervention. In this thesis, we present a such a program and general methods that solve a variety of knowledge discovery problems in GGP. Our main contributions are methods for the automatic construction of heuristic evaluation functions, the automated discovery of game structures, a system for proving properties of games, and symmetry detection and exploitation for general games.:1. Introduction 2. Preliminaries 3. Components of Fluxplayer 4. Game Tree Search 5. Generating State Evaluation Functions 6. Distance Estimates for Fluents and States 7. Proving Properties of Games 8. Symmetry Detection 9. Related Work 10. Discussion