. Planning is investigated in an area where classical Strips- like approaches usually fail. The application domain is therapy (i.e. repair) for complex dynamic processes. The peculiarities of this domain are discussed in some detail for convincingly developing the characteristics of the inductive planning approach presented. Plans are intended to be run for process therapy. Thus, plans are programs. Because of the unavoidable vagueness and uncertainty of information about complex dynamic processes in the case of disturbance, therapy plan generation turns out to be inductive program synthesis. There is developed a graph-theoretically based approach to inductive therapy plan generation. This approach is investigated from the inductive inference perspective. Particular emphasis is put on consistent and incremental learning of therapy plans. Basic application scenarios are developed and compared to each other. The inductive inference approach is invoked to develop and investigate a couple...

The present paper is based on an implemented planning system running in Quintus Prolog under SUN/OS on Sparc Stations. This one has been developed to compete another system previously implemented in Allegro Common Lisp. There have been three essentially different prototypical applications for generating technical therapy plans: a flood prevention system, a chemical fibre production installation, and a ballast tank system for off-shore platforms. This work on planning has been embedded in a comprehensive approach towards knowledgebased process supervision and control within the joint project Wiscon which has been funded by the German Federal Ministry for Research and Technology under grant no. 413--4001--01 IW 204 B. The paper developes a collection of planning algorithms lucidly derived by formalizing algorithmic ideas and heuristics within the logic programming paradigm. This is based on the second author's student's project work.

The problems and solutions adressed in this paper belong to the area of knowledge-based process supervision and control. The focus is on the interaction between a therapy planning and execution system, on the one hand, and a simulation module, on the other hand. Generating therapy plans for seriously disturbed dynamic processes inevitably needs at least approximate knowledge about the future. At a first glance, there is a simple scenario of interaction where a therapy planning system is asking a simulation system for the expected values of certain process parameters within some time interval of interest. A closer look exhibits a number of difficulties. The key problem is that the intermediate results of planning are essential for the work of the simulation module. As a consequence, there has to be a proper interaction of these knowledge processing components. This requires extra efforts of control. The interaction itself is becoming subject to in-depth investigations. There is a tradeo...

The focus of this paper is on the introduction of some modal temporal logic and its application to knowledge processing under the particularly difficult circumstances of disturbed complex dynamic processes. The development of this logic has been driven by the overall aproach to knowledge-based process supervision and control. Within this comprehensive approach, therapy control is playing a crucial role. The therapy control concept comprises the generation of therapy or repair plans, plan execution and constraint monitoring and includes plan revision. The incompleteness and vagueness of available information about an accidentally disturbed process requires a particularly tailored logical approach. The basic concepts are introduced and exploited for developing a collection of planning algorithms properly exceeding previous approaches. 1 Introduction The present paper focusses on a modal temporal logic and its use for planning in complex dynamic environments. Although there are implemen...

. The present paper deals with planning in complex dynamic environments, more specifically, with therapy plan generation for dynamic processes like in chemical industries, for instance. The focus of the investigation is on several closely related phenomena of monotonicity. They have been discovered when developing and implementing novel planning concepts under conditions where Strips-like approaches do not suffice. There is a quite comprehensive approach to knowledge-based process supervision and control sketched as the background of the present investigation. The underlying representation of dynamic knowledge and an integrated simulation module are of a particular interest. Plan generation depends essentially on communication with these components. Under the quite difficult circumstances of a disturbed dynamic process, the generation of therapy plans is inevitably hypothetical. Backtracking during plan generation and plan revision during plan execution are unavoidable and extremely t...

: There is developed an approach to therapy plan generation for complex dynamic processes. This approach has a firm theoretical basis in graph theory using several concepts of hierarchically structured graphs. The underlying theoretical concepts are of a particular importance for developing, investigating, and implementing several heuristics intended to improve the quality of planning algorithms. It is the authors' key intention to explain in detail the way from formal concepts to original algorithmic solutions. The approach is implemented both in Allegro Common LISP and in Quintus Prolog under SUN/OS on SPARC stations. In addition to the running plan generation algorithms interacting with several other components, there are particular tools supporting research work like a plan tracer, e.g. 1 Motivation and Introduction The present paper deals with problems of therapy plan generation for complex dynamic systems. It reports about a couple of theoretical approaches underlying implement...

Introduction and Motivation The intention of the present paper is to clarify that planning under certain circumstances is clearly learning. The necessary assumptions are, first, a well-specified problem domain of planning and, second, a formally well-based and algorithmically clear approach to plan generation 1 . The main result will be a better understanding of the essentials of plan generation under the specified circumstances. Understanding the hypothetical character of generated plans has a serious impact on evaluating and using planning results properly. Furthermore, this approach allows for the development of a couple of novel planning algorithms as refinements of former solutions. As a side effect, the area of machine learning may benefit from exciting stimulations due to the peculiarities of the underlying planning domain. Among others, there is a refinement of former consistency concepts derived from those peculiaritie

. The focus of the present paper is on knowledge representation for innovative knowledge processing under especially difficult circumstances. Typical applications are diagnosis and therapy of disturbed dynamic processes like in chemical industries. The process dynamics need to be represented appropriately to support knowledge processing like therapy plan generation, e.g. The present approach has been implemented and tested successfully. 1 Introduction In the present short paper, technicalities are suppressed as much as possible. Emphasis is put on theory, though there are implemented two planning systems under SUN/OS on Sparc Stations which nicely demonstrate the feasibility of our concepts. This is a paper on Knowledge Representation for Innovative Knowledge Processing under especially difficult circumstances. The feasibility of the approach has been demonstrated. Recent publications reporting about several aspects of this knowledge representation approach are (Jantke and Arnold, 199...

The present publication reports a collection of activities in machine learning research focussing on essential problems of learning from incomplete information. Learning is a central issue of artificial intelligence both in theory and in applications. Under realistic conditions, one can rarely expect complete information about phenomena to be learned. Consequently, learning from incomplete information is central to learning, in general. Inductive inference is the very theory of learning from usually incomplete information. It makes substantial problems explicit and provides a huge collection of results about the power and limitations of inductive learning procedures. The chapters of the present report deal with inductive inference problems varying from fundamentals to applications. Theoretical questions addressed are those for the amount of memory required during learning, for the appropriate information necessary and sufficient for successful learning, and for the ability to recognize...

Abstract not found