Towell, G. G. & Shavlik, J. W. (1993), `The extraction of refined rules from knowledge-based neural networks', Machine Learning 13(1), 17--101.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....allowing them to dynamically modify their learning or pattern classification behavior in real time. The CMT framework may also be used to facilitate network transfer. Network transfer refers to the reuse of neural network parameter values in order to improve the training of new neural networks [1, 17, 19]. Research in network transfer has shown that the use of parameters from existing networks can accelerate the training, and improve the accuracy, of new neural networks [16] Finally, the CMT framework may be used to allow for the implementation of distributed neural networks by providing a means ....

G. G. Towell and J. W. Shavlik. The extraction of refined rules from knowledge-based neural networks. Machine Learning, 13(1):71--101, 1991.

....are developed [15] 16] which can be divided into two groups in general. The first one is architecture analysis based approaches that regard rule extraction as a search process that maps the architecture of the mined neural network to a set of rules. Examples are Fu [17] Towell and Shavlik [18], Sestito and Dillon [19] Setiono [20] Krishnan et al. 21] and Tsukimoto [22] s works. The second one is function analysis based approaches that do not disassemble the architecture of the trained neural networks. Instead, those approaches regard the network as an entity and try to extract ....

G. G. Towell, J. W. Shavlik, The Extraction of Refined Rules from Knowledge-based Neural Networks. Machine Learning, 13(1): 71-101, 1993.

....to new subsets. This is continued until all the subsets are searched. Then, the rules which are over general are removed. For large number of links, finding all subsets becomes computationally very expensive. So the method can only be used for small networks. The basic algorithm is given in [18] and [1] They both establish a limit on the size of the subsets. This will create problems for real world applications. Some real world applications may require large number of antecedents in the rules extracted. 2.3 RULENEG RULENEG was developed by E. Pop [13] an adaptation of the PAC ....

G. G. Towell and J. W. Shavlik. The extraction of refined rules from knowledge-based neural networks. Machine Learning, 13(1):71--101, 1993.

....rule extraction as a search process that maps the architecture of a trained neural network to a set of rules. Examples are as follows. Fu [11] extracts rules for each output unit through searching for subsets of connections whose summed weights exceed the bias of the unit. Towell and Shavlik [42] cluster similar weights in knowledge based artificial neural networks to construct equivalent classes and then extract MOFN rules. Sestito and Dillon [35] generate conjunctive rules using a multilayered network to measure the closeness between inputs and using an inhibitory single layered network ....

G. Towell and J. W. Shavlik, The extraction of refined rules from knowledge-based neural networks, Machine Learning 13 (1993): 71-101.

....allowing them to dynamically modify their learning or pattern classification behavior in real time. The CMT framework may also be used to facilitate network transfer. Network transfer refers to the reuse of neural network parameter values in order to improve the training of new neural networks [1, 17, 19]. Research in network transfer has shown that the use of parameter values from existing networks can accelerate the training, and improve the accuracy, of new neural networks [16] Finally, the CMT framework may be used to allow for the implementation of distributed neural networks by providing a ....

G. G. Towell and J. W. Shavlik. The extraction of refined rules from knowledge-based neural networks. Machine Learning, 13(1):71--101, 1991.

....If, however, the set of rules is correct, but represents only a subset of the set of answers of the network, then the extraction is sound but incomplete. A.S. d Avila Garcez et al. Artificial Intelligence 125 (2001) 155 207 159 pedagogical approaches, while the Subset [10] the MofN [35], the Rulex [3] and Setiono s proposal [29,30] are decompositional methods (see [2] for a comprehensive survey) In the CIL 2 P system, after learning takes place, the network N encodes a knowledge P # that contains the background knowledge P complemented or even revised by the knowledge ....

....to search for subsets of weights of each neuron in the hidden and output layers of N , such that the neurons input potential exceeds its threshold. Each subset that satisfies the above condition is written as a rule. One of the most interesting decompositional methods is the MofN technique [35]. Based on the Subset method, it uses weights clustering and pruning in order to facilitate the extraction of rules. It also generates a smaller number of rules, by taking advantage of the MofNrepresentation, in which m(A 1 , A n ) # A indicates that if m of (A 1 , A n ) are true then A ....

[Article contains additional citation context not shown here]

G.G. Towell, J.W. Shavlik, The extraction of refined rules from knowledge based neural networks, Machine Learning 13 (1993).

....knowledge based classifier, linear programming, quadratic programming 1 Introduction Support vector machine classifiers [15, 1, 12] generate separating planes or surfaces by training on labeled data, that is data for which the class of each point is given. Knowledge based classifiers [13, 14] on the other hand utilize prior knowledge, e.g. an expert s experience such as a doctor s knowledge in diagnosing a certain disease. Recently [7] a precise incorporation of prior knowledge into a linear support vector machine classifier was achieved by placing nonempty polyhedral sets ....

G. G. Towell and J. W. Shavlik. The extraction of refined rules from knowledge-based neural networks. Machine Learning, 13:71--101, 1993.

....overview of the bases of ANN knowledge extraction under the form of logical functions. ANN design relations are established. 1. INTRODUCTION During the last decade considerable effort has been dedicated to write and read symbolic information into and from artificial neural networks (ANNs) [1, 2, 4, 10 14, 16 17, 22 23, 26, 29, 30, 32, 36 38]. The motivation has been multifold. Primarily, ANNs have shown a very good ability to represent empirical knowledge , as the one contained in a set of examples, but the information is expressed in a sub symbolic form i.e. in the structure, weights and biases of a trained ANN, not directly ....

....extraction of the knowledge contained in an ANN allows the portability of the information to other systems, in both symbolic (AI) and sub symbolic (ANN) forms. There are good reasons to consider that Hybrid Learning (HL) Systems able to exploit simultaneously theoretical and empirical data [12 14, 16, 23, 31, 36], would be more efficient than each of the Explanation Based Learning (EBL) systems that Cristea Alexandra, Cristea P. Okamoto T. 2 use only theoretical knowledge in symbolical form, or Empirical Learning (EL) systems handling the knowledge in the ostensive empirical form, working ....

[Article contains additional citation context not shown here]

Towell, G. G. and Shavlik, J. W., "The extraction of refined rules from knowledge based neural networks", Machine learning, vol. 131, pp. 71-101, 1993.

.... reduces the noise, the RNN training becomes more effective, and the symbolic input facilitates the 1 For example, x(t) x(t 1) x(t 2) x(t N 1) form the inputs for a delay embedding of the previous N values of a series [62, 52] 2 Rules can also be extracted from feedforward networks [25, 41, 56, 4, 49, 29], and other types of recurrent neural networks [20] however the recurrent network approach and deterministic finite state automata extraction seem particularly suitable for a time series problem. 2 extraction of rules from the trained networks. Furthermore, it can be argued that the ....

G. Towell and J. Shavlik. The extraction of refined rules from knowledge based neural networks. Machine Learning, 131:71--101, 1993.

....That, in itself, is extremely surprising and can only be achieved by a vanishingly small subset of all possible architectures. Furthermore and Green does not seem to be aware of the research in this area there are a wide range of techniques to extract high level rules from neural networks (Towell Shavlik, 1993). In fact, in cases in which this can be done, one can argue without too much difficulty that the system is following a rule, even if that rule is not implemented as it would have been in a symbolic system. When this is done, groups of units acquire the semantics that boxes and labels have in ....

Towell, G. & Shavlik, J. (1993). The extraction of refined rules from knowledge-based neural networks. Machine Learning, 13:1, 71-101.

....specifies a predicted value for the goal (or class) attribute. This work addresses the well known classification task, widely investigated in the KDD literature. A number of algorithms for extracting rules from neural networks have been proposed in the literature see e.g. 5] 6] 7] 8] [9], 10] 11] 12] 13] In general these algorithms assume that the trained neural network has undergone some postprocessing specifically designed for facilitating the task of rule extraction. For instance, typically the trained network has to be (sometimes drastically) pruned, to reduce the ....

....algorithm. Towell Shavlik [24] developed the SUBSet al..gorithm, based on the decompositional technique, which was very computationally expensive and produced rules with a large number of conditions in their antecedents. In order to improve their method, they developed a new algorithm called MofN [9]. Two other systems using the decompositional technique are RULENET [10] which has the limitation of being oriented for a specific application domain, and RULEX [11] which works with a CEBP (Constrained Error Back Propagation) network. The pedagogical technique views the ANN as an opaque ....

G.G. Towell, and J.W. Shavlik. The Extraction of Refined Rules from Knowledge-Based Neural Networks. Machine Learning, v. 31, n. 1, p. 71-101, 1993.

....high noise and significant non stationarity. In this paper, the noisy times series prediction problem considered is the prediction of foreign exchange rates. A brief overview of foreign exchange rates is presented in the next section. 2 Rules can also be extracted from feedforward networks [25, 41, 56, 4, 49, 29], and other types of recurrent neural networks [20] however the recurrent network approach and deterministic finite state automata extraction seem particularly suitable for a time series problem. 3 1.2 Foreign Exchange Rates The foreign exchange market as of 1997 is the world s largest market, ....

G. Towell and J. Shavlik. The extraction of refined rules from knowledge based neural networks. Machine Learning, 131:71--101, 1993.

....describes the format of the extracted rules. Currently there exist rule knowledge extraction techniques that extract rules in various formats including propositional rules [9] 10[11] fuzzy rules [12] 13] scientific laws [14] finite state automata [15] decision trees [16] and m of n rules [17]. The rule quality criterion is assessed via four characteristics, viz. a) rule accuracy, the extent to which the rule set is able to classify a set of previously unseen examples from the problem domain; b) rule fidelity, the extent to which the extracted rules mimic the behaviour of the ....

.... n) The modules associated with rule simplification are, at worst, polynomial in the number of rules, O(lc 2 ) RULEX is therefore computationally efficient and has some significant advantages over rule extraction algorithms that rely on a (potentially exponential) search and test strategy [10][17]. Thus the use of RULEX to include an explanation facility adds little in the way of overhead to the neural network learning phase. e) Portability RULEX is non portable having been specifically designed to work with local cluster (LC) neural networks. This means that it cannot be used as a ....

G. Towell & J. Shavlik, The Extraction of Refined Rules From Knowledge Based Neural Networks, Machine Learning, Vol 131 (1993), 71-101.

.... [13] Actually hybrid systems have a long story, dating not far from the revival of neural networks (see for instance [14] or [15] or even [16] for a non learning task) and constitute a huge family of learning machines that present many distinguishing features (see for example [17,18,19] KBANN [20] could constitute a paradigmatic and elementary example for this paper. Here the C4.5 [21] or similar [22] mechanisms for symbolically inducing decision trees from data are enriched with neural network facilities for refining decision rules, after a natural mapping of trees into feedforward neural ....

G. Towell and J. Shavlik, "The extraction of refined rules from knowledge based neural networks", Machine learning, vol. 131, pp. 71-101, 1993.

....nodes and weights) Pedagogical methods view neural networks as blackboxes, and use some machine learning algorithm for deriving rules which explain the network input output behavior. For feedforward networks, knowledge has typically been extracted in the form of Boolean and fuzzy ifthen clauses [7, 18]; excellent overviews of the current state of the art can be found in [1, 2] For recurrent networks, finite state automata have been the main paradigm of temporal symbolic knowl 1 2 3 4 5 6 7 8 9 10 Figure 1: Example of DFA: Shown is a minimal, randomly generated DFA with 10 states and 2 input ....

G. G. Towell and J. W. Shavlik, "The extraction of refined rules from knowledge-based neural networks," Machine Learning, vol. 13, no. 1, pp. 71--101, 1993.

....via human is the bottleneck in knowledge acquisition (Boose, 1986) inefficient machine learning is the bottleneck in automated knowledge acquisition. 4 Machine learning can be a continual process, as in people, for revising outdated theories in knowledge based systems (Ourston Mooney, 1990; Towell Shavlik, 1993; Brunk Pazzani, 1995) Many believe that we are poised once again for a radical shift in the way we learn and work, and in the amount of new knowledge we will acquire. The coming age of high performance network computing, and widely available data highways will transform the information age ....

Towell, G. & Shavlik, J. (1993). The extraction of refined rules from knowledgebased neural networks. Machine Learning, 13, 71--101.

....sets, support vector machine classifier, knowledge based classifier, linear programming 1. INTRODUCTION Support vector machines (SVMs) have played a major role in classification problems [24, 3, 12, 1, 13, 5, 6] However un like other classification tools such as knowledge based neural networks [21, 22, 17, 7], little work [20] has gone into incor porating prior knowledge into support vector machines. In this work we present a novel approach to incorporating prior knowledge in the form of polyhedral knowledge sets in the input space of the given data. These knowledge sets, which can be as simple as ....

G. G. Towell and J. W. Shavlik. The extraction of refined rules from knowledge-based neural networks. Machine Learning, 13:71 101, 1993.

No context found.

Towell, G. G. & Shavlik, J. W. (1993), `The extraction of refined rules from knowledge-based neural networks', Machine Learning 13(1), 17--101.

No context found.

Towell G., and Shavlik J. The extraction of refined rules from knowledge based neural networks. Machine Learning, 1993, 13(1): 71-101.

No context found.

G. Towell and J. W. Shavlik, The extraction of refined rules from knowledge-based neural networks, Machine Learning 13 (1993): 71-101.

No context found.

G. Towell and J. Shavlik. The extraction of refined rules from knowledge-based neural networks. Machine Learning, 13(1):71--101, 1993.

No context found.

Towell G., and Shavlik J. The extraction of refined rules from knowledge based neural networks. Machine Learning, 1993, 13(1): 71-101.

No context found.

Towell, G., Shavlik, J.W.: The Extraction of Refined Rules from Knowledge - Based Neural Networks. Machine-Learning, vol.13 (1993)

No context found.

G. Towell and J. W. Shavlik, The extraction of refined rules from knowledge-based neural networks, Machine Learning 13 (1993): 71-101.

No context found.

Towell, G and Shavlik, J "The extraction of refined rules from knowledge based neural networks" Machine Learning Vol 13 No 1 (1993) pp 71-101

First 50 documents

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