David W. Opitz and Jude W. Shavlik. Using genetic search to refine knowledge-based neural networks. In William W. Cohen and Haym Hirsh, editors, Machine Learning, Proceedings of the Eleventh International Conference, pages 208--216, New Brunswick, NJ, 10.-13. July 1994. Morgan Kaufmann, San Mateo, CA. ga94aOpitz.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....Okuno, Taku, 70] Oliker, S. 852, 853] Oliver, I. M. 777] Olmez, T. 71] Omatsu, S. 444] Omatu, Sigeru, 122, 410, 502, 709, 710] Omatu, S. 92, 302] Onami, Saizo, 122] Onami, S. 92, 302] O Neill, A. W. 851] Oosthuizen, G. Deon, 854] Ootani, M. 490] Opitz, David W. [72] Ornes, C. 536] Ortega, J. 153] Osmera, Pavel, 309] Ostrowski, Tomasz, 204, 300, 306, 855] Owechko, Y. 73] Oyro, G. 274] Page, W. C. 61] Pagliarini, Luigi, 296] Pai, G. A. Vijayalakshmi, 541] Pal, Sankar K. 130, 136, 213] Palagi, P. M. 858] Pan, Z. 490] Pao, Y. H. ....

....Sere, Kaisa, 148, 318, 430] Sergeev, S. A. 399] Serra, R. 649, 915] Sette, S. 400] Shaheen, Samir I. 507] Shamir, N. 916] Shams, S. 73] Shang, Yi, 402] Sharman, K. C. 179] Sharman, Ken C. 443] Sharman, Ken, 494] Sharp, David H. 599, 600, 601] Shavlik, Jude W. [72] Sheble, Gerald B. 59] Shi, Y. 312] Shibata, Takanori, 257, 700, 701, 702, 703, 704, 705, 707, 708] Shimohara, Katsunori, 88] Shimojima, Joji, 569] Shimojima, Koji, 147, 294, 310] Shin, Jin Ho, 187] Shine, J. A. 182] Shirao, Yoshiaki, 47] Shonkwiler, Ronald, 917] Siemon, ....

[Article contains additional citation context not shown here]

David W. Opitz and Jude W. Shavlik. Using genetic search to refine knowledge-based neural networks. In William W. Cohen and Haym Hirsh, editors, Machine Learning, Proceedings of the Eleventh International Conference, pages 208--216, New Brunswick, NJ, 10.-13. July 1994. Morgan Kaufmann, San Mateo, CA. ga94aOpitz.

.... [14] and are suitable for difficult problems e.g. problems with very large search spaces and strong nonlinearity, such as the problem of finding good network topologies [15] Although there has been many projects on using genetic algorithms for evolving neural network topologies [15] 16] [17], these projects do not address the issue of extracting comprehensible rules from the evolved network topologies. To the best of our knowledge, this work is the first to combine the idea of using a genetic algorithm to evolve a network topology with the idea of extracting rules from a neural ....

....interconnection(s) inserting some node(s) and its(their) corresponding interconnection(s) and randomly selecting a node or interconnection and its weight. The use of evolutionary algorithms to optimize neural network topologies is not restricted to the classification task. For instance, [17] also used a genetic algorithm to optimize the topology of a neural network, but in their work the network corresponds to a domain knowledge theory (a set of previously known rules) rather than being a network trained from the data. The aim of this work is theory revision, rather than ....

J. W. Shavlik & D. W. Optiz, Using Genetic Search to Refine Knowledge-Based Neural Networks, 11 International Conference of Machine Learning-1994.

....5 Knowledge Refinement As research continues on the problem of using ML for knowledge acquisition, we will develop more guided approaches than the weak search methods. One step that has already been taken in this direction is that of automatically refining incorrect or partial domain knowledge [4, 11, 12, 15, 16, 20, 21, 22, 26, 30, 31, 32, 33, 34, 47, 50]. Even if we do not have a fully satisfactory set of rules for solving a problem, our learning algorithms can still benefit from the incomplete knowledge we do have. Knowledge refinement systems such as those cited are often able to use partial knowledge to produce better solutions to real world ....

D.W. Opitz and J.W. Shavlik. Using genetic search to refine knowledge-based neural networks. In Machine Learning: Proc. 11th Int'l Conf., pages 208--216, New Brunswick, NJ, 1994. Morgan Kaufmann.

....led to the implementation of TB RBF. Other workers investigated methods initializing neural networks (feedforward with sigmoids) using production rules. The idea was to implement domain knowledge, specified by a human expert, in a network. This was done, for instance, by [36] and by [16] 27] [28] describe techniques that further improve the rule based architecture of the neural network. Generally speaking, all these approaches can use the scenario from this paper where the knowledge is induced from the data by machine learning rather than elicited from an expert. The specificity of ....

Opitz, D.W. and Shavlik, J.W. (1994). Using Genetic Search to Refine Knowledge-Based Neural Networks. Proceedings of the 11th International Conference on Machine Learning, New Brunswick, NJ, Morgan Kaufmann

....and database, contributed by M. Noordewier and J. Shavlik to the UCI repository (Murphy Aha, 1992) have become popular for testing systems that integrate empirical and analytical learning (Hirsh Japkowicz, 1994; Koppel, Feldman, Segre, 1994b; Mahoney Mooney, 1994, 1993; Norton, 1994; Opitz Shavlik, 1994; Ortega, 1994; Ourston, 1991; Towell, Shavlik, Noordewier, 1990; Shavlik, Towell, Noordewier, 1992) The original domain theory, as usually interpreted, is overly specific in that it classifies all of the promoter sequences in the database as negative instances. Since the database consists of ....

Opitz, D. W., & Shavlik, J. W. (1994). Using genetic search to refine knowledge-based neural networks. In Proceedings of the Eleventh International Conference on Machine Learning, pp. 208--216 New Brunswick, NJ.

....folds. Therefore it is desirable to have each network use the same validation set. 3.2 Creating and Crossing Over Knowledge Based Neural Networks Steps 1 and 2a in Table 1 specify that new networks need to be created. The algorithm we use for generating these new networks is the Regent algorithm (Opitz Shavlik, 1994). Regent uses genetic algorithms to search through the space of possible neural network topologies. Regent is specifically designed for KNNs, though it applies to standard neural networks as well. Before presenting the exact details of these steps, we discuss (a) how we generate KNNs, and (b) ....

....segment of DNA nucleotides (about 100 elements long) and the task is learn to predict if this DNA subsequence contains a biologically important site. Each of these domains is accompanied by a set of approximately correct rules describing what is currently known about the task (see Opitz, 1995, or Opitz and Shavlik, 1994, for more details) The DNA domains are available at the University of Wisconsin Machine Learning (UW ML) site via the World Wide Web (ftp: ftp.cs.wisc.edu machine learning shavlik group datasets ) or anonymous ftp (ftp.cs.wisc.edu, then cd to machine learning shavlikgroup datasets) Due to ....

[Article contains additional citation context not shown here]

Opitz, D. & Shavlik, J. (1994). Using genetic search to refine knowledge-based neural networks. In Proceedings of the Eleventh International Conference on Machine Learning, (pp. 208--216), New Brunswick, NJ. Morgan Kaufmann.

....hidden units to the weak portions of the network, then repeat as long as improvement is detected. We successfully extended this idea by using genetic operators (specialized versions of crossover and mutation) to create new networks from the current population of candidate network topologies [11]. Finally, we found it useful to create a set of knowledge based networks whose outputs are averaged (weighted) when 3 categorizing new examples [13] we search for a set of highly accurate networks whose errors are not highly correlated. Recently, we extended our approach to the reinforcement ....

D. Opitz and J. Shavlik, "Using genetic search to refine knowledge-based neural networks," in Proceedings of the Eleventh International Conference on Machine Learning, (New Brunswick, NJ), pp. 208--216, Morgan Kaufmann, July 1994.

....the outputs generates a good composite model (Clemen, 1989) we include the predicted accuracy in our weights since one should believe accurate models more than inaccurate ones. 4 Experimental Study The genetic algorithm we use for generating new network topologies is the Regent algorithm (Opitz and Shavlik, 1994). Regent uses genetic algorithms to search through the space of knowledge based neural network (KNN) topologies. KNNs are networks whose topologies are determined as a result of the direct mapping of a set of background rules that represent what we currently know about our task. Kbann (Towell and ....

.... from the Human Genome Project that aid in locating genes in DNA sequences (recognizing promoters, splice junctions, and ribosome binding sites RBS) Each of these domains is accompanied by a set of approximately correct rules describing what is currently known about the task (see Opitz, 1995 or Opitz and Shavlik, 1994 for more details) Our experiments measure the test set error of Addemup on these tasks. Each ensemble consists of 20 networks, and the Regent and Addemup algorithms considered 250 networks during their genetic search. Table 2a presents the results from the case where the learners randomly ....

Opitz, D. and Shavlik, J. (1994). Using genetic search to refine knowledge-based neural networks. In Proceedings of the Eleventh International Conference on Machine Learning, pages 208--216, New Brunswick, NJ. Morgan Kaufmann.

No context found.

D. W. Opitz and J. W. Shavlik. Using Genetic Search to Refine KnowledgeBased Neural Networks. In Machine Learning: Proceedings of the Eleventh International Conference, New Brunswick, NJ, July 1994. Morgan Kaufmann.

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