## Fachbereich Informatik

### Refine

#### Year of publication

- 1996 (32) (remove)

#### Language

- English (32) (remove)

#### Keywords

- AG-RESY (4)
- COMOKIT (3)
- PARO (3)
- Case-Based Reasoning (2)
- CoMo-Kit (2)
- Fallbasiertes Planen (2)
- SKALP (2)
- case-based planning (2)
- Abstraction (1)
- General Knowledge (1)

We present a similarity criterion based on feature weighting. Feature weights are recomputed dynamically according to the performance of cases during problem solving episodes. We will also present a novel algorithm to analyze and explain the performance of the retrieved cases and to determine the features whose weights need to be recomputed. We will perform experiments and show that the integration in a feature weighting model of our similarity criterion with our analysis algorithm improves the adaptability of the retrieved cases by converging to best weights for the features over a period of multiple problem solving episodes.

Planning for manufacturing workpieces is a complex task that requires the interaction of a domain-specific reasoner and a generic planning mechanism. In this paper we present an architecture for organizing the case base that is based on the information provided by a generic problem solver. A retrieval procedure is then presented that uses the information provided by the domain-specific reasoner in order to improve the accuracy of the cases retrieved. However, it is not realistic to suppose that the case retrieved will entirely fit into the new problem. We present a replay procedure to obtain a partial solution that replays not only the valid decisions taken for solving the case, but also justifications of rejected decisions made during the problem solving process. As a result, those completion alternatives of the partial solution are discarded that are already known to be invalid from the case.

Complete Eager Replay
(1996)

We present an algorithm for completely replaying previous problem solving experiences for plan-space planners. In our approach not only the solution trace is replayed, but also the explanations of failed attempts made by the first-principle planner. In this way, the capability of refitting previous solutions into new problems is improved.

This paper addresses the role of abstraction in case-based reasoning. We develop a general framework for reusing cases at several levels of abstraction, which is particularly suited for describing and analyzing existing and designing new approaches of this kind. We show that in synthetic tasks (e.g. configuration, design, and planning), abstraction can be successfully used to improve the efficiency of similarity assessment, retrieval, and adaptation. Furthermore, a case-based planning system, called Paris, is described and analyzed in detail using this framework. An empirical study done with Paris demonstrates significant advantages concerning retrieval and adaptation efficiency as well as flexibility of adaptation. Finally, we show how other approaches from the literature can be classified according to the developed framework.

This paper is to present a new algorithm, called KNNcost, for learning feature weights for CBR systems used for classification. Unlike algorithms known so far, KNNcost considers the profits of a correct and the cost of a wrong decision. The need for this algorithm is motivated from two real-world applications, where cost and profits of decisions play a major role. We introduce a representation of accuracy, cost and profits of decisions and define the decision cost of a classification system. To compare accuracy optimization with cost optimization, we tested KNNacc against KNNcost. The first one optimizes classification accuracy with a conjugate gradient algorithm. The second one optimizes the decision cost of the CBR system, respecting cost and profits of the classifications. We present experiments with these two algorithms in a real application to demonstrate the usefulness of our approach.

When problems are solved through reasoning from cases, the primary kind of knowledge is contained in the specific cases which are stored in the case base. However, in many situations additional background-knowledge is required to cope with the requirements of an application. We describe an approach to integrate such general knowledge into the reasoning process in a way that it complements the knowledge contained in the cases. This general knowledge itself is not sufficient to perform any kind of model-based problem solving, but it is required to interpret the available cases appropriately. Background knowledge is expressed by two different kinds of rules that both must be formalized by the knowledge engineer: Completion rules describe how to infer additional features out of known features of an old case or the current query case. Adaptation rules describe how an old case can be adapted to fit the current query. This paper shows how these kinds of rules can be integrated into an object-oriented case representation.

We present an approach to systematically describing case-based reasoning systems bydifferent kinds of criteria. One main requirement was the practical relevance of these criteria and their usability for real-life applications. We report on the results we achieved from a case study carried out in the INRECA1 Esprit project.

We present a novel approach to classification, based on a tight coupling of instancebased learning and a genetic algorithm. In contrast to the usual instance-based learning setting, we do not rely on (parts of) the given training set as the basis of a nearestneighbor classifier, but we try to employ artificially generated instances as concept prototypes. The extremely hard problem of finding an appropriate set of concept prototypes is tackled by a genetic search procedure with the classification accuracy on the given training set as evaluation criterion for the genetic fitness measure. Experiments with artificial datasets show that - due to the ability to find concise and accurate concept descriptions that contain few, but typical instances - this classification approach is considerably robust against noise, untypical training instances and irrelevant attributes. These favorable (theoretical) properties are corroborated using a number of hard real-world classification problems.

This article will discuss a qualitative, topological and robust world-modelling technique with special regard to navigation-tasks for mobile robots operating in unknownenvironments. As a central aspect, the reliability regarding error-tolerance and stability will be emphasized. Benefits and problems involved in exploration, as well as in navigation tasks, are discussed. The proposed method demands very low constraints for the kind and quality of the employed sensors as well as for the kinematic precision of the utilized mobile platform. Hard real-time constraints can be handled due to the low computational complexity. The principal discussions are supported by real-world experiments with the mobile robot

In this paper we describe how explicit models of software or knowledge engineering processes can be used to guide and control the distributed development of complex systems. The paper focuses on techniques which automatically infer dependencies between decisions from a process model and methods which allow to integrate planning and execution steps. Managing dependencies between decisions is a basis for improving the traceability of develop- ment processes. Switching between planning and execution of subprocesses is an inherent need in the development of complex systems. The paper concludes with a description of the CoMo-Kit system which implements the technolo- gies mentioned above and which uses WWW technology to coordinate development processes. An on-line demonstration of the system can be found via the CoMo-Kit homepage:

Paris (Plan Abstraction and Refinement in an Integrated System) [4, 2] is a domain independent case-based planning system which allows the flexible reuse of planning cases by abstraction and refinement. This approach is mainly inspired by the observation that reuse of plans must not be restricted to a single description level. In domains with a high variation in the problems, the reuse of past solutions must be achieved at various levels of abstraction.

EADOCS (Expert Assisted Design of Composite Structures) is the implementation of a multi-level approach to conceptual design. Constraint-, case- and rule-based reasoning techniques are applied in different design phases to assemble and adapt designs at increasing levels of detail. This paper describes a strategic approach to decomposition, formulation of target design problems, and incremental retrieval and adaptation. Design problems considered, cannot be decomposed dynamically into tractable subproblems. Design cases are retrieved for requirements and preferences on both functionality and the solution. Cases are adapted in three phases: adaptation, modification and optimisation.

Representations of activities dealing with the development or maintenance of software are called software process models. Process models allow for communication, reasoning, guidance, improvement, and automation. Two approaches for building, instantiating, and managing processes, namely CoMo-Kit and MVP-E, are combined to build a more powerful one. CoMo-Kit is based on AI/KE technology; it was developed for supporting complex design processes and is not specialized to software development processes. MVP-E is a process-sensitive software engineering environment for modeling and analyzing software development processes, and guides software developers. Additionally, it provides services to establish and run measurement programmes in software organizations. Because both approaches were developed completely independently major integration efforts are to be made to combine their both advantages. This paper concentrates on the resulting language concepts and their operationalization necessary for building automated process support.

In this report we give an overview of the development of our new Waldmeisterprover for equational theories. We elaborate a systematic stepwise design process, startingwith the inference system for unfailing Knuth - Bendix completion and ending up with animplementation which avoids the main diseases today's provers suffer from: overindulgencein time and space.Our design process is based on a logical three - level system model consisting of basicoperations for inference step execution, aggregated inference machine, and overall controlstrategy. Careful analysis of the inference system for unfailing completion has revealed thecrucial points responsible for time and space consumption. For the low level of our model,we introduce specialized data structures and algorithms speeding up the running system andcutting it down in size - both by one order of magnitude compared with standard techniques.Flexible control of the mid - level aggregation inside the resulting prover is made possible by acorresponding set of parameters. Experimental analysis shows that this flexibility is a pointof high importance. We go on with some implementation guidelines we have found valuablein the field of deduction.The resulting new prover shows that our design approach is promising. We compare oursystem's throughput with that of an established system and finally demonstrate how twovery hard problems could be solved by Waldmeister.

A combination of a state-based formalism and a temporal logic is proposed to get an expressive language for various descriptions of reactive systems. Thereby it is possible to use a model as well as a property oriented specification style in one description. The descriptions considered here are those of the environment, the specification, and the design of a reactive system. It is possible to express e.g. the requirements of a reactive system by states and transitions between them together with further temporal formulas restricting the behaviors of the statecharts. It is shown, how this combined formalism can be used: The specification of a small example is given and a designed controller is proven correct with respect to this specification. The combination of the langugages is based on giving a temporal semantics of a state-based formalism (statecharts) using a temporal logic (TLA).

We present first steps towards fully automated deduction that merely requiresthe user to submit proof problems and pick up results. Essentially, this necessi-tates the automation of the crucial step in the use of a deduction system, namelychoosing and configuring an appropriate search-guiding heuristic. Furthermore,we motivate why learning capabilities are pivotal for satisfactory performance.The infrastructure for automating both the selection of a heuristic and integra-tion of learning are provided in form of an environment embedding the "core"deduction system.We have conducted a case study in connection with a deduction system basedon condensed detachment. Our experiments with a fully automated deductionsystem 'AutoCoDe' have produced remarkable results. We substantiate Au-toCoDe's encouraging achievements with a comparison with the renowned the-orem prover Otter. AutoCoDe outperforms Otter even when assuming veryfavorable conditions for Otter.

Evolving Combinators
(1996)

One of the many abilities that distinguish a mathematician from an auto-mated deduction system is to be able to offer appropriate expressions based onintuition and experience that are substituted for existentially quantified variablesso as to simplify the problem at hand substantially. We propose to simulate thisability with a technique called genetic programming for use in automated deduc-tion. We apply this approach to problems of combinatory logic. Our experimen-tal results show that the approach is viable and actually produces very promisingresults. A comparison with the renowned theorem prover Otter underlines theachievements.This work was supported by the Deutsche Forschungsgemeinschaft (DFG).

We present a concept for an automated theorem prover that employs a searchcontrol based on ideas from several areas of artificial intelligence (AI). The combi-nation of case-based reasoning, several similarity concepts, a cooperation conceptof distributed AI and reactive planning enables a system using our concept tolearn form previous successful proof attempts. In a kind of bootstrapping processeasy problems are used to solve more and more complicated ones.We provide case studies from two domains of interest in pure equationaltheorem proving taken from the TPTP library. These case studies show thatan instantiation of our architecture achieves a high grade of automation andoutperforms state-of-the-art conventional theorem provers.

We investigate the usage of so-called inference rights. We point out the prob-lems arising from the inflexibility of existing approaches to heuristically controlthe search of automated deduction systems, and we propose the application ofinference rights that are well-suited for controlling the search more flexibly. More-over, inference rights allow for a mechanism of "partial forgetting" of facts thatis not realizable in the most controlling aproaches. We study theoretical founda-tions of inference rights as well as the integration of inference rights into alreadyexisting inference systems. Furthermore, we present possibilities to control suchmodified inference systems in order to gain efficiency. Finally, we report onexperimental results obtained in the area of condensed detachment.The author was supported by the Deutsche Forschungsgemeinschaft (DFG).

Planning for realistic problems in a static and deterministic environment with complete information faces exponential search spaces and, more often than not, should produce plans comprehensible for the user. This article introduces new planning strategies inspired by proof planning examples in order to tackle the search-space-problem and the structured-plan-problem. Island planning and refinement as well as subproblem refinement are integrated into a general planning framework and some exemplary control knowledge suitable for proof planning is given.

Load balancing is one of the central problems that have to be solved in parallel computation. Here, the problem of distributed, dynamic load balancing for massive parallelism is addressed. A new local method, which realizes a physical analogy to equilibrating liquids in multi-dimensional tori or hypercubes, is presented. It is especially suited for communication mechanisms with low set-up to transfer ratio occurring in tightly-coupled or SIMD systems. By successive shifting single load elements to the direct neighbors, the load is automatically transferred to lightly loaded processors. Compared to former methods, the proposed Liquid model has two main advantages. First, the task of load sharing is combined with the task of load balancing, where the former has priority. This property is valuable in many applications and important for highly dynamic load distribution. Second, the Liquid model has high efficiency. Asymptotically, it needs O(D . K . Ldiff ) load transfers to reach the balanced state in a D-dimensional torus with K processors per dimension and a maximum initial load difference of Ldiff . The Liquid model clearly outperforms an earlier load balancing approach, the nearest-neighbor-averaging. Besides a survey of related research, analytical results within a formal framework are derived. These results are validated by worst-case simulations in one-and two-dimensional tori with up to two thousand processors.

One of the many features needed to support the activities of autonomous systems is the ability of motion planning. It enables robots to move in their environment securely and to accomplish given tasks. Unfortunately, the control loop comprising sensing, planning, and acting has not yet been closed for robots in dynamic environments. One reason involves the long execution times of the motion planning component. A solution for this problem is offered by the use of highly computational parallelism. Thus, an important task is the parallelization of existing motion planning algorithms for robots so that they are suitable for highly computational parallelism. In several cases, completely new algorithms have to be designed, so that a parallelization is feasible. In this survey, we review recent approaches to motion planning using parallel computation.

We have presented a novel approach to parallel motion planning for robot manipulators in 3D workspaces. The approach is based on arandomized parallel search algorithm and focuses on solving the path planning problem for industrial robot arms working in a reasonably cluttered workspace. The path planning system works in the discretized con guration space, which needs not to be represented explicitly. The parallel search is conducted by a number of rule-based sequential search processes, which work to find a path connecting the initial con guration to the goal via a number of randomly generated subgoal con gurations. Since the planning performs only on-line collision tests with proper proximity information without using pre-computed information, the approach is suitable for planning problems with multirobot or dynamic environments. The implementation has been carried outontheparallel virtual machine (PVM) of a cluster of SUN4 workstations and SGI machines. The experimental results have shown that the approach works well for a 6-dof robot arm in a reasonably cluttered environment, and that parallel computation increases the e ciency of motion planning signi cantly.

The paper presents a novel approach to parallel motion planning for robot manipulators in 3D workspaces. The approach is based on a randomized parallel search algorithm and focuses on solving the path planning problem for industrial robot arms working in a reasonably cluttered workspace. The path planning system works in the discretized configuration space which needs not to be represented explicitly. The parallel search is conducted by a number of rule-based sequential search processes, which work to nd a path connecting the initial configuration to the goal via a number of randomly generated subgoal configurations. Since the planning performs only on-line collision tests with proper proximity information without using pre-computed information, the approach is suitable for planning problems with multirobot or dynamic environments. The implementation has been carried out on the parallel virtual machine (PVM) of a cluster of SUN4 workstations and SGI machines. The experimental results have shown that the approach works well for a 6-dof robot arm in a reasonably cluttered environment, and that parallel computation increases the efficiency of motion planning significantly.

This document offers a concise introduction to the Goal Question Metric Paradigm (GQM Paradigm), and surveys research on applying and extending the GQM Paradigm. We describe the GQM Paradigm in terms of its basic principles, techniques for structuring GQM-related documents, and methods for performing tasks of planning and implementing a measurement program based on GQM. We also survey prototype software tools that support applying the GQM Paradigm in various ways. An annotated bibliography lists sources that document experience gained while using the GQM Paradigm and offer in-depth information about the GQM Paradigm.

In recent years, Smolyak quadrature rules (also called hyperbolic cross points or sparse grids) have gained interest as a possible competitor to number theoretic quadratures for high dimensional problems. A standard way of comparing the quality of multivariate quadrature formulas
consists in computing their \(L_2\)-discrepancy. Especially for larger dimensions, such computations are a highly complex task. In this paper we develop a fast recursive algorithm for computing the \(L_2\)-discrepancy (and related quality measures) of general Smolyak quadratures. We carry out numerical comparisons between the discrepancies of certain Smolyak rules, Hammersley and Monte Carlo sequences.

A notion of discrepancy is introduced, which represents the integration error on spaces of \(r\)-smooth periodic functions. It generalizes the diaphony and constitutes a periodic counterpart to the classical \(L_2\)-discrepancy as weil as \(r\)-smooth versions of it introduced recently by Paskov [Pas93]. Based on previous work [FH96], we develop an efficient algorithm for computing periodic discrepancies for quadrature formulas possessing certain tensor product structures, in particular, for Smolyak quadrature rules (also called sparse grid methods). Furthermore, fast algorithms of computing periodic discrepancies for lattice rules can easily be derived from well-known properties of lattices. On this basis we carry out numerical comparisons of discrepancies between Smolyak and lattice rules.

Software development organizations measure their real-world processes, products, and resources to achieve the goal of improving their practices. Accurate and useful measurement relies on explicit models of the real-world processes, products, and resources. These explicit models assist with planning measurement, interpreting data, and assisting developers with their work. However, little work has been done on the joint use of measurem(int and process technologies. We hypothesize that it is possible to integrate measurement and process technologies in a way that supports automation of measurement-based feedback. Automated support for measurementbased feedback means that software developers and maintainers are provided with on-line, detailed information about their work. This type of automated support is expected to help software professionals gain intellectual control over their software projects. The dissertation offers three major contributions. First, an integrated measurement and
process modeling framework was constructed. This framework establishes the necessary foundation for integrating measurement and process technologies in a way that will permit automation. Second, a process-centered software engineering environment was developed to support measurement-based feedback. This system provides personnel with information about the tasks expected of them based on an integrated set of measurement and process views. Third, a set of assumptions and requirements about that system were examined in a controlled experiment. The experiment compared the use of different levels of automation to evaluate the acceptance and effectiveness of measurement-based feedback.

Quasi-Monte Carlo Radiosity
(1996)

The problem of global illumination in computer graphics is described by a second kind Fredholm integral equation. Due to the complexity of this equation, Monte Carlo methods provide an interesting tool for approximating
solutions to this transport equation. For the case of the radiosity equation, we present the deterministic method of quasi-rondom walks. This method very efficiently uses low discrepancy sequences for integrating the Neumann series and consistently outperforms stochastic techniques. The method of quasi-random walks also is applicable to transport problems in settings other
than computer graphics.

The calculation of form factors is an important problem in computing the global illumination in the radiosity setting. Closed form solutions often are only available for objects without obstruction and are very hard to calculate. Using Monte Carlo integration and ray tracing provides a fast and elegant tool for the estimation of the form factors. In this paper we show, that using deterministic low discrepancy sample points is superior to random sampling, resulting in an acceleration of more than half an order of magnitude.