Abstract
PDF (2337 K)
E-mail Article
Add to my Quick Links
Cited By in Scopus (0)
doi:10.1016/S0004-3702(96)00024-0 
Copyright © 1996 Published by Elsevier Science B.V.
Using action-based hierarchies for real-time diagnosis
David Ash
, 
, * and Barbara Hayes-Roth
Available online 16 February 1999.
Abstract
An intelligent agent diagnoses perceived problems so that it can respond to them appropriately. Basically, the agent performs a series of tests whose results discriminate among competing hypotheses. Given a specific diagnosis, the agent performs the associated action. Using the traditional information-theoretic heuristic to order diagnostic tests in a decision tree, the agent can maximize the information obtained from each successive test and thereby minimize the average time (number of tests) required to complete a diagnosis and perform the appropriate action. However, in real-time domains, even the optimal sequence of tests cannot always be performed in the time available. Nonetheless, the agent must respond. For agents operating in real-time domains, we propose an alternative action-based approach in which: (a) each node in the diagnosis tree is augmented to include an ordered set of actions, each of which has positive utility for all of its children in the tree; and (b) the tree is structured to maximize the expected utility of the action available at each node. Upon perceiving a problem, the agent works its way through the tree, performing tests that discriminate among successively smaller subsets of potential faults. When a deadline occurs, the agent performs the best available action associated with the most specific node it has reached so far. Although the action-based approach does not minimize the time required to complete a specific diagnosis, it provides positive utility responses, with step-wise improvements in expected utility, throughout the diagnosis process. We present theoretical and empirical results contrasting the advantages and disadvantages of the information-theoretic and action-based approaches.
Author Keywords: Reactive planning; Decision trees; Diagnosis; Real-time planning; Heuristics
References
[1]. D. Ash, Diagnosis using action-based hierarchies for optimal real-time performance. In: Ph.D. dissertation, Computer Science Department, Stanford University, Stanford, CA (1994).
[2]. D. Ash, G. Gold, A. Seiver and B. Hayes-Roth, Guaranteeing real-time response with limited resources. J. Artif. Intell. Med. 5 (1992), pp. 49–66.
[3]. I. Bratko, I. Mozeti
and N. Lavra
, Automatic synthesis and compression of cardiological knowledge. In: J.E. Hayes, D. Michie and J. Richards, Editors, Machine Intelligence 11, Oxford University Press, Oxford (1987), pp. 435–454.
[4]. L. Breiman, J. Friedman, R. Olshen and C. Stone. In: Classification and Regression Trees, Wadsworth & Brooks, Monterey, CA (1984).
[5]. R. Brooks, A robust layered control system for a mobile robot. IEEE J. Robot. Automat. (March 1986), pp. 14–23. View Record in Scopus | Cited By in Scopus (1444)
[6]. J. Cheng, U. Fayyad, K. Irani and Z. Qian, Improved decision trees: a generalized version of ID3. In: Proceedings Fifth International Conference on Machine Learning (1988), pp. 100–108.
[7]. L. Chrisman and R. Simmons, Sensible planning: focusing perceptual attention. In: Proceedings AAAI-91 (1991), pp. 756–761.
[8]. W. Clancey, Heuristic classification. Artif. Intell. 27 (1985), pp. 289–350. Abstract | 
PDF (3396 K)
| View Record in Scopus | Cited By in Scopus (164)
[9]. P. Clark and T. Niblett, The CN2 induction algorithm. Mach. Learn. 3 (1989), pp. 261–284.
[10]. J. Clarke, M. Niv, B. Webber, K. Fisherkeller, D. Southerland and R. Ryack, TraumAID: a decision aid for managing trauma at various levels of resources. In: Proceedings Thirteenth Annual Symposium on Computer Applications in Medical Care (1989).
[11]. G. Cooper, The computational complexity of probabilistic inference on Bayesian belief networks. Artif. Intell. 42 (1990), pp. 353–405.
[12]. V. Dabija, Deciding whether to plan to react. In: Ph.D. dissertation, Computer Science Department, Stanford University, Stanford, CA (1994).
[13]. R. Davis and W. Hamscher, Model-based reasoning: troubleshooting. In: H. Shrobe, Editor, Exploring Artificial Intelligence: Survey Talks from the National Conference on Artificial Intelligence (1988), pp. 297–346.
[14]. T. Dean and M. Boddy, An analysis of time-dependent planning. In: Proceedings AAAI-88 (1988), pp. 49–54.
[15]. T. Dean, L. Kaelbling, J. Kirman and A. Nicholson, Planning with deadlines in stochastic domains. In: Proceedings AAAI-93 (1993), pp. 574–579. View Record in Scopus | Cited By in Scopus (20)
[16]. J. de Kleer and B. Williams, Diagnosing multiple faults. Artif. Intell. 32 (1987), pp. 97–130. Abstract | PDF (1784 K)
| View Record in Scopus | Cited By in Scopus (316)
[17]. M. Factor, The process trellis software architecture for parallel real-time monitors. In: Ph.D. dissertation, Yale University, New Haven, CT (1990).
[18]. L. Fagan, VM: representing time-dependent relations in a medical setting. In: Ph.D. dissertation, Computer Science Department, Stanford University, Stanford, CA (1980).
[19]. U. Fayyad, On the induction of decision trees for multiple concept learning. In: Ph.D. dissertation, EECS Department, University of Michigan, Ann Arbor, MI (1991).
[20]. U. Fayyad and K. Irani, The attribute selection problem in decision tree generation. In: Proceedings AAAI-92 (1992), pp. 104–110. View Record in Scopus | Cited By in Scopus (27)
[21]. R. Fikes, P. Hart andized robot plans. Artif. Intell. 3 (1972), pp. 251–288. Abstract | PDF (2353 K)
| View Record in Scopus | Cited By in Scopus (90)
[22]. K. Forbus, Qualitative process theory. Artif. Intell. 24 (1984), pp. 85–168. Abstract | PDF (3948 K)
| View Record in Scopus | Cited By in Scopus (317)
[23]. M. Ginsburg, Universal planning: an (almost) universally bad idea. AI Mag. 4 (1989), pp. 40–44.
[24]. J. Hendler and A. Agrawala, Mission critical planning: AI on the MARUTI real-time operating system. In: K. Sycara, Editor, Proceedings Workshop on Innovative Approaches to Planning, Scheduling, and Control (1990), pp. 77–84.
[25]. M. Henrion, Search-based methods to bound diagnostic probabilities in very large belief nets. In: Proceedings Seventh International Conference on Uncertainty in AI (1991), pp. 142–150.
[26]. P. Hoel. In: Introduction to Mathematical Statistics, Wiley, New York (1947).
[27]. E. Horvitz, Reasoning about beliefs and actions under computational resource constraints. In: Proceedings Third Workshop on Uncertainty in AI (1987).
[28]. L. Kaelbling, An architecture for intelligent reactive systems. In: M. Georgeff and A. Lansky, Editors, Reasoning about Actions and Plans (1987), pp. 395–410.
[29]. L. Kaelbling, Goals as parallel program specifications. In: Proceedings AAAI-88 (1988).
[30]. L. Kaelbling, Specifying complex behavior for computer agents. In: Proceedings Workshop on Innovative Approaches to Planning, Scheduling and Control (1990), pp. 433–438.
[31]. B. Kuipers, Qualitative simulation. Artif. Intell. 29 (1986), pp. 289–338. Abstract | PDF (2296 K)
| View Record in Scopus | Cited By in Scopus (236)
[32]. S. Murthy, S. Kasif, S. Salzberg and R. Beigel, A randomized induction of oblique decision trees. In: Proceedings AAAI-93 (1993), pp. 322–327. View Record in Scopus | Cited By in Scopus (15)
[33]. R. Quinlan, Inductive inference as a tool for the construction of high-performance programs. In: R.S. Michalski, T.M. Mitchell and J. Carbonell, Editors, Machine Learning, Tioga, Palo Alto, CA (1983).
[34]. R. Quinlan, Probabilistic decision trees. In: Y. Kodratoff and R. Michalski, Editors, Machine Learning: An Artificial Intelligence Approach 3, Morgan Kaufmann, San Mateo, CA (1990).
[35]. R. Reiter, A theory of diagnosis from first principles. Artif. Intell. 32 (1987), pp. 57–95. Abstract | PDF (1710 K)
| View Record in Scopus | Cited By in Scopus (417)
[36]. S. Rosenschien, Synthesizing information-tracking automata from environment descriptions. In: Proceedings First International Conference on Principles of Knowledge Representation and Reasoning (1989).
[37]. S. Russell and E. Wefald, Principles of metareasoning. Artif. Intell. 49 (1991), pp. 361–395. Abstract | PDF (2023 K)
| View Record in Scopus | Cited By in Scopus (36)
[38]. G. Rutledge, Dynamic selection of models under time constraints. In: Proceedings Second Annual Conference on AI Simulation and Planning in High Autonomy Systems (1991), pp. 60–67.
[39]. G. Rutledge, G. Thomsen, B. Farr, M. Tovar, J. Polaschek, I. Beinlich, L. Sheiner and L. Fagan, The design and implementation of a ventilator-management advisor. Artif. Intell. Med. (1993).
[40]. M. Schoppers, Universal plans for reactive robots in unpredictable environments. In: Proceedings IJCAI-87 (1987).
[41]. E. Shortliffe. In: MYCIN: Computer-Based Consultations in Medical Therapeutics, Elsevier, New York (1976).
[42]. D. Tong, Weaning patients from mechanical ventilation: a knowledge-based system approach. Comput. Meth. Programs Biomed. 35 (1991), pp. 267–278. Abstract | Article | PDF (889 K)
| View Record in Scopus | Cited By in Scopus (14)
[43]. S. Uckun, B. Dawant and D. Lindstrom, Model-based reasoning in intensive-care monitoring: the YAQ approach. Artif. Intell. Med. (1993).
[44]. L. Widman, A model-based approach to the diagnosis of the cardiac arrhythmias. Artif. Intell. Med. 4 (1991), pp. 1–19.
[45]. J. Xu, S. Hyman and P. King, Knowledge-based flash evoked potential recognition system. Artif. Intell. Med. 4 (1992), pp. 93–109. Abstract | PDF (1261 K)
| View Record in Scopus | Cited By in Scopus (1)
* Present address: Brightware Inc., 90 Park Avenue, Suite 1600, New York, NY 10016, USA.
