K. Forbus and B. Falkenhainer. Self-Explanatory Simulations: Scaling Up to Large Models. In Proceedings of AAAI-92, page To Appear, 1992.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....1) 1 MO , MO : Monotonic function I , I : Influenced Figure 4. A generated causal network. 4 Explanation Generation Once the causal network is obtained, we can generate explanations of the system s behavior by propagating qualitative values in the net work [Forbus and Falkenhainer, 1990] [Forbus and Falkenhainer, 1992]. By tracing a loop in the causal network, we obtain an explanation: transition from search to attracted transition from attracted to trace transition from trace to transport In the generated explanatiom the reasoner finds a positive feedback among the parameters. This means that there is a ....

K. D. Forbus and B. Falkenhainer. Self-explanatory simulations: Scaling up to large models. In Proc. of AAAI-92, pages 685 690, 1992.

....hand, requires only a qualitative model, and uses the model, not to correct the cases, but to develop indices indicating which cases are most relevant to the current situations. A number of people in the qualitative modeling community have added quantitative information to qualitative models. In [ 4 ] , Forbus and Falkenhainer use qualitative envisionment combined with numerical modeling information; this approach can work well, but requires that one provide both the qualitative and numerical models up front. Kay and Kuipers, in [ 7 ] take the approach of quantitatively constraining function ....

K. Forbus, B. Falkenhainer: "Self-Explanatory Simulations: Scaling up to large models," Proceedings: 10th National Conference on Artificial Intelligence, 1992

....as for example in Q3 [2] Using recently improved techniques of interval constraint propagation ( 10, 5] it is possible to obtain results with sufficient precision to be of practical use. It is also possible to use qualitative models for generating numerical simulation models, such as SIMGEN [7]. For differential equations, phase space techniques ( 13, 14] can offer more accuracy. However, these techniques require significantly more than purely qualitative knowledge, and thus may not be realistic in many cases. 3 Memory based Reasoning: lacking structure In its most general form, ....

K. Forbus, B. Falkenhainer: "Self-Explanatory Simulations: Scaling up to large models," Proceedings of the 10th National Conference of the AAAI, AAAI/MIT Press, 1992

....and Daniel S. Weld 3 Department of Computer Science and Engineering, FR 35 University of Washington Seattle, Washington 98195 franz, adam, weld cs.washington.edu Abstract We present Pika, an implemented self explanatory simulator that is more than 5000 times faster than SimGen Mk2 [ Forbus and Falkenhainer, 1992 ] the previous state of the art. Like SimGen, Pika automatically prepares and runs a numeric simulation of a physical device specified as a particular instantiation of a general domain theory, and it is capable of explaining its reasoning and the simulated behavior. Unlike SimGen, Pika s ....

.... has been much recent work on automating their construction (e.g. Yang, 1992, Rosenberg and Karnopp, 1983, Abelson and Sussman, 1987, Palmer and Cremer, 1992 ] To this, the Qualitative Physics community has contributed the idea of a self explanatory simulator [ Forbus and Falkenhainer, 1990, Forbus and Falkenhainer, 1992 ] When using such a system, a person need only specify the basic entities, quantities, and equations governing the system to be simulated. From these, the program automatically prepares and runs a numeric simulation. It also keeps a record of its reasoning so it can explain the simulated ....

[Article contains additional citation context not shown here]

K. Forbus and B. Falkenhainer. Self-Explanatory Simulations: Scaling Up to Large Models. In Proceedings of AAAI-92, page To Appear, 1992.

....algebra to build a 3D spatial reasoning system, and a 2D shape modeler. KEYWORDS: Qualitative Reasoning; Spatial Reasoning; Shape; Approximation; Hybrid Qualitative Reasoning 1 Introduction Artificial Intelligence has long been concerned with the interaction between abstraction and detail ([4], 14] At one end of the abstraction spectrum are the traditional numerical approaches, often called quantitative, and at the other end are a number of knowledge representation paradigms, many of which This work was done while the author was at the Dept. of Computer Science and Engineering, ....

Forbus, Kenneth D., and Brian Falkenhainer (1992). Self-explanatory simulations: scaling up to large models, AAAI-92, p.6875-690

....conditions. The architecture shown is implemented in the DME system (Iwasaki Low, 1991) The basic technology required for rationale by explanation is model formulation assistance (Falkenhainer Forbus, 1991; Iwasaki Low, 1991; Nayak, 1992; Palmer Cramer, 1991) and explainable simulation (Falkenhainer Forbus, 1992; Gruber Gautier, 1992) It only works in domains where formal simulation models exist, and is only practical if the model formulation process can be made efficient. Research on both fronts is progressing rapidly. In any case, the potential payoffs to the user are significant. Not only is ....

Falkenhainer, B., & Forbus, K. (1992). Self-explanatory simulations : Scaling up to large models. Proceedings of the Tenth National Conference on Artificial Intelligence, San Jose, CA, pages 685-690, AAAI press / MIT press.

....behaviors of interest, and assumed operating environment, and the system generates a low level equation model for simulation. It can then explain the predicted behavior in terms of the high level descriptions. DME s explanation approach is most similar to the approach taken in the SIMGEN systems [3,7]. Like DME, these systems use a compositional modeling method of model formulation, perform numeric simulation, and generate causal explanations. However, they use a qualitative model for explanation and a parallel quantitative model for simulation. DME generates qualitative explanations from the ....

....event rules. The intent is to facilitate the development of model libraries by domain specialists not trained in qualitative modeling. In any case, useful model libraries are difficult to build and new development tools will be required to achieve this goal. The successor to SIMGEN, SIMGEN.MK2 [3], addresses the efficiency and scale problems with a compilation approach. SIMGEN.MK2 precomputes the conditions for state transitions from an analysis of the qualitative model, without using a global envisionment. Nevertheless, the time required by SIMGEN.MK2 to produce a self explanatory ....

B. Falkenhainer & K. Forbus. Self-explanatory simulations: Scaling up to large models. Proceedings of the Tenth National Conference on Artificial Intelligence, San Jose, CA, pages 685-690. AAAI press / The MIT press, 1992.

....of Q2 and NSim can be intersected and must be consistent, otherwise the behaviour can be refused. SQSim can make more precise predictions than each of the methods alone could do. However, SQSim inherits the limitations of QSim as discussed above. Based on QPT (Forbus 1984) SimGen is presented in (Forbus Falkenhainer 1990; 1992). SimGen performs a qualitative analysis on a qualitative model to predict possible qualitative behaviours. For each behaviour a numeric model can be inferred for a numeric simulation run. Pika (Amador, Finkelstein, Weld 1993) is another self explanatory simulator, similar to SimGen. In contrast ....

Forbus, K. D., and Falkenhainer, B. 1992. Self-explanatory simulations: Scaling up to large models. In Proc. 10th National Conf. on Artificial Intelligence (AAAI-92), 685-- 690. Menlo Park, CA: AAAI Press/The MIT Press.

....not scale. Qualitative models have been used to generate explanations in tutoring and training systems [10,21] DME s explanation system can also generate explanations on such models (using QSIM [15] for simulation) Qualitative models have known limitations of scale. Work on the SIMGEN systems [5,9] was the first to achieve the effect of qualitative explanation using numerical simulation models. SIMGEN also uses a compositional modeling approach, and explains simulation data using the derivation of equations from model fragments. The SIMGEN strategy is to build parallel qualitative and ....

B. Falkenhainer & K. Forbus. Self-explanatory simula - tions: Scaling up to large models. AAAI-92, pp. 685-690, 1992.

....or quantitative non causal equations to encode physics based models. The simulator produces qualitative or quantitative self explanatory simulations. It is a Lisp based interpreter and allows dynamic assembly. The qualitative simulations are produced with QSIM [55] 3.2. 2 SIMGEN SIMGEN [34, 35, 36] is a semi qualitative self explanatory simulation compiler. It uses the modeling language of the Qualitative Process Engine (QPE) 31] which is directly based on the Qualitative Process Theory [29] SIMGEN is currently used to produce simulations for online virtual laboratories [71] SIMGEN s ....

....models that will be used during the simulation. Therefore, the resulting simulation code is almost as good as if it had been hand crafted. However, this proved impractical because of the exponential explosion of qualitative states. Total envisionment was dropped in the subsequent versions Mk2 [35] and Mk3 [36] Mk3 runs in polynomial time. SIMGEN supports all the features of compositional modeling. It compiles a scenario into a stand alone C simulation program. In particular all the possible view instantiation conditions are hard coded in C. The system assembly cannot be modified at ....

[Article contains additional citation context not shown here]

Kenneth D. Forbus and Brian Falkenhainer. Self-explanatory simulations: Scaling up to large models. In Proc. 10th National Conf. on Artificial Intelligence (AAAI-92), pages 380--387. AAAI Press/The MIT Press, 1990.

....by another program and describes algorithms for detecting assumption violations and other problems that might lead to low quality or unreliable simulation results, but does not discuss what an automated design system should do after detecting a low quality result. Forbus and Falkenhainer 1990, Forbus and Falkenhainer 1992, Forbus and Falkenhainer 1995] discuss the use of qualitative simulation to check the quality of numerical simulation results, but here strategies for dealing with modeling failures in an automated design system are also not discussed. A great deal of work has been done in the area of numerical ....

Kenneth D. Forbus and Brian Falkenhainer. Selfexplanatory simulations: Scaling up to large models. In Proceedings, Tenth National Conference on Artificial Intelligence, San Jose, CA, 1992.

....by another program and describes algorithms for detecting assumption violations and other problems that might lead to low quality or unreliable simulation results, but does not discuss what an automated design system should do after detecting a low quality result. Forbus and Falkenhainer 1990, Forbus and Falkenhainer 1992, Forbus and Falkenhainer 1995 ] discuss the use of qualitative simulation to check the quality of numerical simulation results, but here strategies for dealing with modeling failures in an automated design system are also not discussed. A number of research efforts have combined AI techniques ....

Kenneth D. Forbus and Brian Falkenhainer. Selfexplanatory simulations: Scaling up to large models. In Proceedings, Tenth National Conference on Artificial Intelligence, San Jose, CA, 1992.

....theories to be tested only under certain operating conditions (e.g. steady state assumptions) ffl The lack of integration between the qualitative models and the quantitative knowledge used by engineers in their domains. The recent development of self explanatory simulator compilers like SIMGEN [21] extends the ways in which qualitative methods can be used in the design process. The main idea behind these systems is the automatic creation of simulators that integrate qualitative and numerical models during simulation. These programs can then be used either as ordinary numerical simulators ....

....and computes the minimal sets of conditions under which a model fragment is active (Fig. 2.2) The results of this step are used to activate qualitative model fragments that are consistent with the input parameters. This analysis corresponds to the first step of the qualitative analysis in SIMGEN [21]. OUZO demonstrates that this type of qualitative analysis is general enough to support typical conceptual design tasks like the design of separation systems that deal with either steady state or macroscopic models for physical systems. 2.3.2.2 Numerical Model Construction Numerical model ....

[Article contains additional citation context not shown here]

Forbus, K. D. and Falkenhainer, B., Self-Explanatory Simulations: Scaling up to large models. AAAI-92, July 1992.

....This is exactly the kind of information that is available in the water supply and many other engineering domains. Until recently, no technique was developed for both supporting model management and coping with incomplete knowledge. Few systems [Forbus and Falkenhainer, 1990; Iwasaki and Low, 1991; Forbus and Falkenhainer, 1992; Amador et al. 1993] have been developed that use compositional modeling techniques, and exploit qualitative models to provide explanations of simulations. They are unable, however, to represent or use semi quantitative information. In order to provide any sort of numeric simulation, they all ....

....simulators that construct numerical simulations and use the qualitative representation to help explain the results. Unlike SQPC, they do not use semi quantitative information. Their predictions are either precise numeric ones, or purely qualitative. SIMGEN [Forbus and Falkenhainer, 1990; Forbus and Falkenhainer, 1992] computes a total envisionment of the scenario and then, for each envisionment state 6 it builds a numerical simulator, monitors the simulation and, at the end of the analysis, interprets numerical results in terms of the envisionment graph. SIMGEN requires precise and complete numerical ....

K. Forbus and B. Falkenhainer. Self-- explanatory simulations: scaling up to large models. In Proc. of the tenth National Conference on Artificial Intelligence, pages 685--690, 1992.

.... boolean Figure 15: Cost to achieve a range of design qualities with 99 confidence. Quality is takeoff mass, normalized by the best takeoff mass found (Figure 11) so quality = 1.01 corresponds to Figure 14. ures to an automated design systems are not discussed. Forbus and Falkenhainer 1990, Forbus and Falkenhainer 1992, Forbus and Falkenhainer 1995] discuss the use of qualitative simulation to check the quality of numerical simulation results, but here strategies for communicating information about modeling failures to an automated design systems are also not discussed. Other automated intelligent controllers ....

Kenneth D. Forbus and Brian Falkenhainer. Selfexplanatory simulations: Scaling up to large models. In Proceedings, Tenth National Conference on Artificial Intelligence, San Jose, CA, 1992.

....and describes algorithms for detecting assumption violations and other problems that might lead to low quality or unreliable simulation results, but strategies for communicating information about modeling failures to an automated design systems are not discussed. Forbus and Falkenhainer 1990, Forbus and Falkenhainer 1992, Forbus and Falkenhainer 1995 ] discuss the use of qualitative simulation to check the quality of numerical simulation results, but here strategies for communicating information about modeling failures to an automated design systems are also not discussed. Other automated intelligent controllers ....

Kenneth D. Forbus and Brian Falkenhainer. Selfexplanatory simulations: Scaling up to large models. In Proceedings, Tenth National Conference on Artificial Intelligence, San Jose, CA, 1992.

....without requiring experimental data as for differential equations. Bobrow [15] identified different qualitative reasoning tasks ; among them are simulation, envisionment (determining all possible behaviours) diagnosis and verification. Even if this approach has limited utility for large models [16] , mainly because of combi natorics, it tends to allow the modelling of phenomena for which the numerical accuracy is not required, or for which behaviour laws are not completely known. Qualitative simulation has been already applied in biology. MOLGEN. 17] supports the qualitative simulation of ....

Forbus, K.D., Falkenhainer, B., "Self-Explanatory Simulations : Scaling up to large models", Proc. of the 10th National Conference on Artificial Intelligence AAAI-92, July 12-16, San Jose, Ca, pp. 685-690, 1992

....the numerical models that currently implement the functional vocabulary. By supporting qualitative reasoning about control systems, we should be able to build programs that provide critiques of a student s design. 2. We are extending the Feedback MiniLab to use self explanatory simulators [10,11] as system simulations. Using self explanatory simulators should both improve the quality of explanations and support more interesting student critiques. For example, if the qualitative analysis predicts that the student s controller won t correct properly The initial version of this system was ....

Forbus, K. and Falkenhainer, B. Self-Explanatory Simulations: Scaling up to large models", Proceedings of AAAI-92, July, 1992. Submitted to AAAI93 Reasoning about

No context found.

K. Forbus and B. Falkenhainer. Self-Explanatory Simulations: Scaling Up to Large Models. In Proceedings of AAAI-92, page To Appear, 1992.

No context found.

Forbus, K.D. & Falkenhainer, B. (1992). Self-explanatory simulations: Scaling up to large models. In Proceedings of AAAI-92, pages 685-690.

No context found.

Forbus, K.D. & Falkenhainer, B. (1992). Self-explanatory simulations: Scaling up to large models. In Proceedings of AAAI-92, pp. 685-690.

No context found.

Forbus, K.D. & Falkenhainer, B. (1992). Self-explanatory simulations: Scaling up to large models. In Proceedings of AAAI-92, pages 685-690.

No context found.

Brian Falkenhainer and Kenneth D. Forbus. Self-Explanatory Simulations: Scaling up to Large Models. AAAI-92 685--690. Also in QR-92 22--35.

No context found.

Brian Falkenhainer and Kenneth D. Forbus. Self-Explanatory Simulations: Scaling up to Large Models. AAAI-92 685--690. Also in QR-92 22--35.

No context found.

Falkenhainer, B., and Forbus, K. D. 1992. SelfExplanatory Simulations: Scaling up to Large Models.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC