Y. Peng and J.A. Reggia. A connectionist model for diagnostic problem solving. IEEE Transactions on Systems, Man and Cybernetics, 1989.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....both time and space exponential in the cluster size that equals the induced width of the network s moral graph. Following Pearl s stochastic simulation algorithms [32] the suitability of Stochastic Local Search (SLS) algorithms for MPE was studied in the context of medical diagnosis applications [33] and more recently in [17] Best First search algorithms were proposed [41] as well as algorithms based on linear programming [37] Various authors have worked on extending some of these algorithms to the task of finding the k most likely explanations [24, 42] 2.6 Bucket and Mini Bucket ....

Y. Peng and J.A. Reggia. A connectionist model for diagnostic problem solving. IEEE Transactions on Systems, Man and Cybernetics, 1989.

....both time and space exponential in the cluster size that equals the induced width of the network s moral graph. Following Pearl s stochastic simulation algorithms [29] the suitability of Stochastic Local Search (SLS) algorithms for MPE was studied in the context of medical diagnosis applications [30] and more recently in [17] Best First search algorithms were proposed [39] as well as algorithms based on linear programming [36] Various authors have worked on extending some of these algorithms to the task of finding the k most likely explanations [23, 40] 2.6 Bucket and Mini Bucket ....

Y. Peng and J.A. Reggia. A connectionist model for diagnostic problem solving. IEEE Transactions on Systems, Man and Cybernetics, 1989.

....and its extension by conditioning and clustering, and his stochastic simulation proposal [9] researchers have investigated various approaches, especially in the context of medical diagnosis. Our work on greedy algorithms can be viewed as an extension of the line of work presented in [10] [11], ranging from two layered networks to general belief networks. More recently, best first search algorithms were proposed [17] as well as algorithms based on linear programming [15] Various other authors have worked on extending some of these algorithms to the task of finding the k most likely ....

Peng, Y., Reggia, J. A., 1989. A Connectionist Model for Diagnostic Problem Solving, In IEEE Transactions on Systems, Man and Cybernetics, Vol. 19, No. 2 March/April.

.... mathematical treatment of its theoretical basis is given in [PR87a] and [P 90] It is an efficient algorithm that requires relatively small amounts of data to operate, and is flexible enough to be used in real world applications under many different computational methods [PR87a, PR87b, PR89, P 90, P 92] The MFD problem solved consisted of ten possible symptoms and fifteen possible disorders. This creates a search space of 2 15 possible diagnoses for any one of the possible 1023 symptom sets. The symptom set with no symptoms was not considered. An exhaustive search for ....

Yun Peng and James A. Reggia. A connectionist model for diagnostic problem solving. IEEE Transactions on Systems, Man, and Cybernetics, 19(2):285--298, 1989. 10

....the most likely disease a patient is suffering from. In decoding, the task is to identify the most likely input message which was transmitted over a noisy channel, given the observed output. Researchers have investigated various approaches to finding the mpe in a belief network. See, e.g. [35, 9, 36, 37]) Recent proposals include best first search algorithms [48] and algorithms based on linear programming [41] The problem is to find x 0 such that P (x 0 ) max x P (x; e) max x Pi i P (x i ; ejx pa i ) where x = x 1 ; xn ) and e is a set of observations, on subsets of the ....

Y. Peng and J.A. Reggia. A connectionist model for diagnostic problem solving. IEEE Transactions on Systems, Man and Cybernetics, 1989.

....well only if the network is sparse enough to allow small cutsets or small clusters. Following Pearl s stochastic simulation algorithms for the MPE task [Pearl, 1988] the suitability of Stochastic Local Search (SLS) algorithms for MPE was studied in the context of medical diagnosis applications [Peng and Reggia, 1989] and more recently in [Kask and Dechter, 1999b] Best first search algorithms were also proposed in [Shimony and Charniak, 1991] as well as algorithms based on linear programming [Santos, 1991] 2 Background 2.1 Notation and definitions Belief Networks provide a formalism for reasoning about ....

Y. Peng and J.A. Reggia. A connectionist model for diagnostic problem solving. IEEE Transactions on Systems, Man and Cybernetics, 1989.

....and flexible simulation environment for neural network research. The activation rules and the computing procedure are implemented with C program functions which are inserted into SNNS. By means of SNNS, the parallel computation of the neural network model is implemented. The two examples used by [7] are borrowed and tested by the model proposed here. We call the two examples as Example 1 and Example 2, respectively. Each of the two examples has a causal network of 10 disorders and 10 manifestations. The details of the two causal networks are given in TABLE 1. For a causal network with n ....

Peng, Y. and Reggia, J., "A Connectionist Model for Diagnostic Problem Solving", IEEE Trans. On Systems, Man and Cybernetics, VOl.19, pages 285298, 1989.

....made a lot of efforts on alleviating the intractable difficulty. Neural network computing, an approach based on highly parallel local computations, is known to be strong in solving computationally difficult tasks. A great deal of work has been made in applying neural networks to diagnostic tasks [1, 2, 3, 4, 5]. Associative networks have long been studied as a knowledge representation in AI. An associative network usually consists of nodes, representing entities such as objects, concepts, and events, and links between the nodes, representing the interrelations or associations between nodes. Their ....

Peng, Y. and Reggia, J., "A Connectionist Model for Diagnostic Problem Solving", IEEE Trans. On Systems, Man and Cybernetics, Vol.19, No.2, pages 285-198, 1989.

....scheduling, involving up to 1600 rows and 105000 columns. Genetic algorithms and simulated annealing algorithms for set covering appear in [Sen93, BC94] Several neural network based algorithms were suggested or developed for problems related to SCP (like scheduling and diagnostic problems, cf. PR89, Jef91, CM91] but to our knowledge no neural network based algorithm for the SCP was actually presented and tested so far. One of the algorithms tested in this work is such a neural net based algorithm that was developed recently. It will be described in detail separately [Gro94] A ....

Y. Peng and J. A. Reggia. A connectionist model for diagnostic problem solving. IEEE Transactions on Systems, Man, and Cybernetics, 19:285, 1989.

....an optimal solution much faster. We investigate this approach for the Most Probable Explanation (MPE) task. It appears in applications such as medical diagnosis, circuit diagnosis, natural language understanding and probabilistic decoding. Some earlier work on MPE can be found in [ Pearl, 1988; Peng and Reggia, 1989; Shimony and Charniak, 1991; Santos, 1991 ] Section 3 presents the relevant algorithms against which we will be comparing. Section 4 describes our branch and bound scheme and its guiding heuristic function. Section 5 presents the empirical evaluations while section 6 provides discussion and ....

Y. Peng and J.A. Reggia. A connectionist model for diagnostic problem solving. IEEE Transactions on Systems, Man and Cybernetics, 1989.

....the cancellation diagnosis can be performed. 1 Introduction Neural network techniques are often used for information processing, such as signal processing. Recent years, many efforts have been made in applying neural networks to a number of AI tasks including diagnostic tasks [Jakubowicz, 1990; Peng, 1989; Goel, 1996; Xu, 1998] There are four types of diagnoses: independent diagnosis, monotonic diagnosis, incompatible diagnosis, and cancellation diagnosis. The diagnostic problems solved by almost all the existing neural network diagnosis models belong to independent diagnosis. The problem solved ....

Peng, Y., and Reggia, J. (1989). A Connectionist Model for Diagnostic Problem Solving, IEEE Trans. On Systems, Man and Cybernetics, 19(2): 285-198.

....reported in this table were run with a population size of 50. # Reasonable estimates were used here because of excessive runtimes. 26 The four bar charts go here. All 4 on a page. 27 7. Conclusions Although the tested search techniques are not applied to real world MFD data (we follow [Peng89] by generating tendency matrix data at random but with certain constraints such as the density) the reliabilities obtained do shed some light on the behavior of these methods. In summary, the following conclusions can be made about the different heuristics and hybrids used, namely, GA , EC , ....

Y. Peng and J.A. Reggia, "A Connectionist Model for Diagnostic Problem Solving," in IEEE Transactions on Systems, Man, and Cybernetics, Vol. SMC-19, No. 2, pp. 285-298, March/April 1989.

....enough to allow small cutsets or small clusters. Following Pearl s stochastic simulation algorithms for the MPE task [ Pearl, 1988 ] the suitability of Stochastic Local Search (SLS) algorithms for MPE was studied in the context of Medical diagnosis applications [ Peng and Reggia, 1986 ] Peng and Reggia, 1989 ] and more recently in [ Kask and Dechter, 1999b ] Best first search algorithms were also proposed [ Shimony and Charniak, 1991 ] as well as algorithms based on linear programming [ Santos, 1991 ] 2 Background 2.1 Notation and definitions Belief Networks provide a formalism for reasoning ....

Y. Peng and J.A. Reggia. A connectionist model for diagnostic problem solving. IEEE Transactions on Systems, Man and Cybernetics, 1989.

....the reasons is the large size of the possible combinations of individual hypotheses whose number is exponentially large, thus diagnostic problem solving becomes combinatorially difficulty. Recent years, the approach on developing neural network diagnosis models has received more and more attention[1,2,3]. For a neural network diagnosis model, generally, it includes three parts: network architecture, activation rules to calculate the node activations, and the network equilibrium which indicates the stop condition of network computing. The activation rules and the network equilibrium vary with ....

....one possible solution to the diagnostic problem. By means of the highly parallel local computations of neural networks, the difficulty mentioned above can be alleviated. For the neural network diagnosis models, the probabilistic causal network is often directly adopted as the network architecture[3,4]. Generally, the disorders in the probabilistic causal network are assumed to be independent of each other, i.e. the disorder independence assumption is one of the basic assumptions for the neural network diagnosis model. With this assumption, the existence of a disorder is assumed not being ....

[Article contains additional citation context not shown here]

Peng, Y., and Reggia, J., "A Connectionist Model for Diagnostic Problem Solving", IEEE Trans. On Systems, Man and Cybernetics, 19(2), pages 285-198, 1989.

....a winner take all competitive behavior that ensures that only one of a set of incompatible nodes may be active when the network reaches an acceptable state. An alternative approach to producing useful competitive behavior is through a technique called competition based spreading activation (CBSA) (Peng Reggia, 1989; Reggia, 1987; Reggia, Peng, Bourret, 1991; Sutton, 1992) In this approach, competing processing units vie for a portion of the fixed amount of activation being output from a common B A C .4 .6 Figure 3.3: Competition through competition based spreading activation. Nodes A and B have no ....

Peng, Y. and J. Reggia (1989). "A Connectionist Model for Diagnostic Problem Solving." IEEE Transactions on Systems, Man, and Cybernetics 19:2, 285--298.

....are identical. In ongoing research, we are studying how such storage can be optimized. 9 A Diagnostic Example We now present an application of our work to diagnostic problems. The form of diagnostic reasoning is based on the parsimonious covering theory developed by Jim Reggia and his colleagues [31]. In their work, we are given a knowledge base consisting of a finite set of physical states D about which the knowledge base contains information, a finite set M of manifestations of these states and some causal information of the form: State d causes manifestation m. The above causal ....

....program. A disjunctive logic program is the same as a logic program with negations in clause bodies; computing the minimal models of such programs can be done using if(P ) We show below a specific knowledge base for diagnosing chemical spills. Example 12 The knowledge base below is taken from [31]. 1. sulfuric acid hydrochloric acid carbonic acid benzenesulfonic acid) ph(acidic) 2. chromogen r23 hydroxyaluminum rubidium) ph(alkaline) 3. petroleum watercolor(black) 4. thioacetamide chromogen r23 hydroxyaluminum) watercolor(red) 5. benzene petroleum benzenesulfonic ....

Y. Peng and J. Reggia. (1989) A Connectionist Model for Diagnostic Problem Solving, IEEE Trans. on Systems, Man and Cybernetics, 19, 2, pps 285--298.

....such as diagnosis and abduction. For example, it can suggest the disease from which a patient suffers given data on clinical findings. Researchers have investigated various approaches to finding the mpe in a belief network. See, e.g. Pearl, 1988; Cooper, 1984; Peng and Reggia, 1986; Peng and Reggia, 1989)) Recent proposals include best first search algorithms (Shimony and Charniack, 1991) and algorithms based on linear programming (Santos, 1991) The problem is to find x 0 such that P (x 0 ) max x Pi i P (x i ; ejx pa i ) where x = x 1 ; x n ) and e is a set of observations. Namely, ....

Y. Peng and J.A. Reggia. A connectionist model for diagnostic problem solving, 1989.

....network diagnosis model, it includes three parts: network architecture, activation rules to calculate the node activations, and the network equilibrium which indicates the stop condition of network computing. The probabilistic causal networks are often directly adopted as the network architectures [4,5]. The activation rules and the network equilibrium vary with different models. Given a set of manifestations, the node activations are calculated by the activation rules repeatedly until the network equilibrium is reached. Thus, a set of disorders will be determined by the activations of disorder ....

....d i (t Gamma 1) j is close to 0 without question. In this case, j d i (t) Gamma d i (t Gamma 1) j 0 stops the computing, and a solution is found. 4 Experiments We have conducted a number of experiments to test the neural network model and also to make comparisons with the model proposed in [5]. All the experiments are performed on a neural network simulator called SNNS(Stuttgart Neural Network Simulator) which provides an efficient and flexible simulation environment for neural network research. The activation rules and the computing procedure are implemented with C program functions ....

[Article contains additional citation context not shown here]

Peng, Y., and Reggia, J., "A Connectionist Model for Diagnostic Problem Solving", IEEE Trans. On Systems, Man and Cybernetics, Vol.19, No.2, pp 285-198, 1989.

....Reggia s multiprocessor scheme. The difference lies in that our model associates a process with each h 2 H e and each d 2 D o . This enables considerably more distribution of processing and allows for exploitation of parallelism with in H e and D o . In the connectionist paradigm, Peng and Reggia [24] describe a competition based neural network for solving a class of problems in medical diagnosis. Like our second network, their network contains connections among the neurons representing h 2 H e and d 2 D o where the connections represent the explanatory relations between H e and D o . In ....

....varied from a few thousand to several hundred thousand depending on the initial conditions of the network. Finally, it is not clear how to accommodate cancellation interactions among elementary hypotheses in a competition based scheme. In fact, none of the competition based schemes described in [24, 34, 35] can handle this type of interaction. This is important because abductive inference is non monotonic in the presence of this interaction. The ability to accommodate cancellation interactions in abduction problems appears to require some kind of additional processing. 5.2 Critique of the Neural ....

Y. Peng and J. Reggia, "A Connectionist Model for Diagnostic Problem Solving," IEEE Trans. Systems Man and Cybernetics, 19:285-289, 1989.

No context found.

Y. Peng and J.A. Reggia. A connectionist model for diagnostic problem solving. IEEE Transactions on Systems, Man and Cybernetics, 1989.

No context found.

Y. Peng and J. Reggia, "A connectionist model for diagnostic problem solving," IEEE Transactions on Systems, Man and Cybernetics 19 (1989): pp. .

No context found.

Y. Peng and J.A. Reggia. A connectionist model for diagnostic problem solving. IEEE Transactions on Systems, Man and Cybernetics, 1989.

No context found.

Y. Peng and J.A. Reggia. A connectionist model for diagnostic problem solving. IEEE Transactions on Systems, Man and Cybernetics, 1989.

No context found.

Y. Peng and J.A. Reggia. A connectionist model for diagnostic problem solving. IEEE Transactions on Systems, Man, and Cybernetics, 19(2):285-- 298, 1989.

No context found.

Y. Peng and J. Reggia, "A connectionist model for diagnostic problem solving," IEEE Transactions on Systems, Man and Cybernetics 19 (1989): pp. .

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