S. F. Smith. A learning system based on genetic adaptive algorithms. Doctoral dissertation, Department of Computer Science. University of Pittsburgh, 1980.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....often means two, so that binary strings are used for the genotypes. There is nothing obligatory in taking a one bit range for each allele but there are theoretical reasons to prefer few alleles at many sites over many alleles at few sites (the arguments have been given by [Holland 1975] p. 71) [Smith 1980](p. 56) and supporting evidence for the correctness of these arguments has been presented by [Schaffer 1984] p. 107) The function v provides a measure of fitness for a given phenotype and (since the programmer must also supply a mapping # from the set of genotypes to the set of phenotypes) ....

S. F. Smith. A Learning System Based on Genetic Adaptive Algorithms. Ph.D. Dissertation, University of Pittsburg, 1980.

....of multiprocessor task scheduling [3] Although our initial results are somewhat surprising, a detailed analysis of the evolutionary dynamics provide an interesting and positive explanation. 2 Related work Early work on variable length representations includes Smith s LS 1 learning system [4], Goldberg s messy GA [5] Koza s genetic programming (GP) 6] 2 Yu et al. and Harvey s SAGA [7] These studies laid much of the groundwork in terms of defining the issues that need to be addressed with variable length representations and exploring potential solutions. Since then, much research ....

....Accordingly, all task sequences in both solutions are valid with respect to G1. With respect to problem G2, all task sequences in the smaller solution are valid, but only the following subset of task sequences (18 out of 52 total) from the longer solution are valid: 14 13] 1 3] 3 2] 1 3 2] [6 4] [4 5] 6 4 5] 7 6] 6 9] 9 8] 7 6 9] 6 9 8] 7 6 9 8] 8 12] 12 11] 11 10] 12 11 10] 4 9] The longer solution is a complete solution to problem G1. It is very specialized for G1, and consequently, noticeably less fit for G2. The shorter solution does not specify a complete solution for ....

[Article contains additional citation context not shown here]

Smith, S.F.: A learning system based on genetic adaptive algorithms. In: PhD thesis, Dept. Computer Science, University of Pittsburgh. (1980)

....a problem unique to genetic programming. It occurs in a wide variety of arbitrary length representations, including neural networks, finite state automata, and rule sets. Indeed, the earliest known report of bloating (and of approaches to deal with it) involves evolving Pitt approach rule systems [1]. However, because GP is the most popular arbitrary length representation technique, the lion s share of papers on the subject have been in the GP literature. As discussed in [2] bloating is a hotly debated topic in GP theory; and there is also presently no silver bullet to deal with it. The ....

Stephen F. Smith. A Learning System Based on Genetic Adaptive Algorithms. PhD thesis, Computer Science Department, University of Pittsburgh, 1980.

.... evolutionary programming has its roots in a form of programming) And in the 1980 s, genetic algorithms were applied to evolving rule based systems such as classifier systems [18] Steve Smith developed one of the first systems applying genetic search to variable length rule based systems in 1980 [41]. But genetic programming represents a major change in paradigm. To start to understand genetic programming, it is perhaps best to look at a restricted example. Assume we are given the following function approximation task, where we wish to approximate a function of the form Genetic ....

S. Smith. A learning systems based on genetic adaptive algorithms. PhD thesis, University of Pittsburgh, 1980.

.... The objective of this section is to give a brief description of the architecture of our learning environment called DELVAUX, and to explain the semantics of Bayesian classification rules that we try to learn (a more detailed discussion of these topics has been given in [EJ93] 1 See [DJ90] and [SM80] for a more detailed discussion of the Pittsburgh approach. 3 2.1 The DELVAUX Learning Environment In general, we assume that we have a significant number of examples. These examples might be part of a database in which we plan to discover interesting relationships between various attributes of ....

Smith, S.F.: "A Learning System Based on Genetic Adaptive Algorithms, Doctoral Thesis, Department of Computer Science, University of Pittsburgh, Pittsburgh 1980.

.... Pittsburgh Versus Michigan Approach There have historically been two approaches to genetic classification, named after the universities at which the approaches originated: the Michigan approach (Booker et al. 1989; Holland, 1986) and the Pittsburgh approach (De Jong et al. 1993; Janikow, 1993; Smith, 1980, 1983) The main property distinguishing the two approaches is whether each population element represents a single classification rule or a set of rules. Although the two approaches have come with other accessories, we will use this single property to define and distinguish them. In the Michigan ....

Smith, S. F. 1980. A learning system based on genetic adaptive algorithms. Doctoral dissertation. University of Pittsburgh, Pittsburgh, Pa.

....of the genetic algorithm. Because only one individual in the population is chosen among many, it is necessary to encode the entire rule set of the system into each individual. This approach was used by S.F. Smith in his LS 1 system, whose most famous application 19 was a Poker playing system [36]. A comparison with the canonical classifier system model, CS 1, can be found in [18] An intuitive argument can be made for the increased problem complexity from using LS 1 as opposed to CS 1 by considering two multiple rule production systems, each which must discover R rules. Let n be the ....

S.F. Smith. A Learning System Based on Genetic Adaptive Algorithms. PhD thesis, University of Pittsburgh, 1980. 75

....search method for a variety of learning tasks. In fact, there has been a good deal of interest in using GAs for machine learning problems [72] Three alternative approaches, in which GAs have been applied to learning processes, have been proposed, the Michigan [73] the Pittsburgh [74], and the Iterative Rule Learning (IRL) 46] approaches. In the rst one, the chromosomes correspond to classi er rules that are evolved as a whole, whereas in the Pittsburgh approach, each chromosome encodes a complete set of classi ers. In the IRL approach each chromosome represents only one ....

Smith, S. F. (1980). A learning system based on genetic adaptive algorithms. Ph. D. Thesis, University of Pittsburgh.

....rules, etc. Over the last few years, these advantages have extended the use of GAs in the development of a wide range of approaches for designing FRBSs [8, 11, 15, 16, 17, 27] Three alternative approaches have been proposed to apply GAs to learning processes: the Michigan [31] the Pittsburgh [37], and the Iterative Rule Learning (IRL) 41] approaches. In the rst one, the chromosomes correspond to classi er rules that are evolved as a whole, whereas in the Pittsburgh approach, each chromosome encodes a complete set of classi ers. In the IRL approach each chromosome represents a single ....

S.F. Smith, A learning system based on genetic adaptive algorithms, Ph. D. Thesis (University of Pittsburgh, 1980).

.... [248] University of New Mexico, 219] University of North Carolina at Chapel Hill, 319] University of North Carolina at Charlotte, 369] University of Oklahoma, 231] University of Paris 7, 171] University of Paris XII, 341] University of Pennsylvania, 159] University of Pittsburgh,[318, 363] University of Pretoria, 386] University of Reading, 268] University of Rochester, 241] Master s theses 13 University of Sao Paulo, 251] University of Sheeld, 215] University of Southern California, 187] University of Stirling, 309] University of Strathclyde, 343] University of ....

....Gerrit Imme, 29] Sekharan, D. Ansa, 134] Self, Steven, 135] Semertzidis, Michael T. 171] Sen, Sandip, 129] Seniw, D. 369] Shu, Lingyan, 330] Shyu, Ming Suen, 39] Siegmund, Frederik, 136] Sittisathanchai, Sinchai, 172] Smith, Robert Elliot, 351, 125] Smith, Stephen F. [363] Sprave, Joachim, 137] Sputek, K. 138] Starkweather, Timothy John, 370] Stevens, J. 139] Swaminathan, R. 30] Syed, Omar, 40] Tackett, Walter Alden, 187] Tadikonda, Satish Kumar, 372] Talbi, El Ghazali, 373] Tanese, Reiko, 374] Taylor, Timothy John, 254] Thangiah, Sam ....

[Article contains additional citation context not shown here]

Stephen F. Smith. A learning system based on genetic adaptive algorithms. PhD thesis, University of Pittsburgh, 1980. (University Microlms No. 81-12638) yMichalewicz92book ga:SFSmithThesis.

....class. Second, in their approach a population consists of a single rule set, called Michigan approach [2] in the genetic algorithm literature, whereas in our rule learning environment members of the population are complete rule sets and not single rules, which is called Pittsburgh approach [16] in the literature. Although the discussion why we prefer to learn fuzzy rules, that rely on multi valued logic and whose inputs are possibilities or probabilities, is beyond the scope of the paper, we still like to give our motivation why we make this kind of rule the focus of our scientific ....

S. F. Smith, "A learning system based on genetic adaptive algorithms," Ph.D. dissertation, Univ. Pittsburgh, 1980.

....and prove a useful resource for researchers. Although the rst classi er system, CS 1, was reported in 1978 [233] the development of LCS was foreshadowed by some of Holland s earlier work [218, 219, 220] dating back as far as 1971. In the early 80 s much progress was in the form of PhD theses [384, 37, 190, 349, 170] (but see also [423, 424] following which LCS papers began to appear steadily in conferences. Later landmarks include the publication of books by Holland in 1986 [231] and Goldberg in 1989 [193] and the rst and second international workshops on LCS, in 1992 and 1999 respectively. As the ....

S. F. Smith. A Learning System Based on Genetic Adaptive Algorithms. PhD thesis, University of Pittsburgh, 1980.

....lies is directly influenced by the representation. It is widely recognised that fixed length character strings pose severe limitations for the solution of particular problems [5] 6] Several representation schema which improve the capabilities of Genetic Algorithms have been presented by Smith [21] Koza [13] and Harvey [8] 9] 10] Smith has provided an earlier example of using variable length strings in representing the genome. Harvey has shown that adaptation of variable length genome improves conventional Genetic Algorithms by changing their finite search space to an open ended search. ....

S.S. Smith. A learning system based on genetic adaptive algorithms, phd. dissertation, university of pittsburgh, 1980.

....suitable representation both capable of gathering the problem characteristics and sutiably representing the potential solutions to it. Classically, two genetic learning approaches, adopted from the field of genetic based machine learning systems, have been used: the Michigan [7, 40] and Pittsburgh [55] approaches. In the Michigan approach, the chromosomes are individual fuzzy rules and the FRB is represented by the entire population. The collection of fuzzy rules is adapted over time using some genetic operators applied at the level of the individual rule. This evolution is guided by a credit ....

S.F. Smith, A Learning System Based on Genetic Adaptive Algorithms, PhD thesis, University of Pittsburgh (1980).

....their algorithms can be adapted and extended for our present purposes. The method of doing so will be here presented; C code for implementing this is available from the author. 2. Examples of Variable length systems VLGs have been proposed for a number of purposes, e.g. Smith s LS 1 classifiers [16], Koza s Genetic Programming [12] Goldberg s Messy GAs [2] Harp and Samad s genetic synthesis of neural network architectures [4] Care needs to be taken that a crossover operation exchanges meaningful building blocks. In the case of LS 1 this is relatively simple, as a genotype is effectively a ....

....of control systems for ill defined domains; such as autonomous robots. The theoretical underpinning for GAs, Holland s Schema Theorem [9, 3] is no longer valid when the genotypes within a population vary in length. Where an analysis for VLGs has been offered, as in Smith s LS 1 classifiers [16] and Koza s genetic programming [12] the analyses offered have not satisfactorily extended the notion of a schema such that schemata are preserved by the genetic operators [5, 7] The conceptual framework of SAGA was introduced in 1991 in order to try to understand the dynamics of a GA when ....

Stephen F. Smith. A Learning System based on Genetic Adaptive Algorithms. PhD thesis, Department of Computer Science, University of Pittsburgh, USA, 1980.

No context found.

S. F. Smith. A learning system based on genetic adaptive algorithms. Doctoral dissertation, Department of Computer Science. University of Pittsburgh, 1980.

No context found.

Smith, S.: A Learning System Based on Genetic Adaptive Algorithms. PhD thesis, Department of Computer Science, University of Pittsburgh (1980)

No context found.

Stephen F. Smith. A Learning System Based on Genetic Adaptive Algorithms. PhD thesis, University of Pittsburgh, 1980.

No context found.

Stephen F. Smith. A Learning System Based on Genetic Adaptive Algorithms. PhD thesis, University of Pittsburgh, 1980.

No context found.

S.F. Smith, A learning system based on genetic adaptive algorithms. Ph. D. Thesis. University of Pittsburgh. 1980.

No context found.

S.F. Smith, A Learning System Based on Genetic Adaptive Algorithms. Ph. D. Thesis. University of Pittsburgh. 1980. Fig. 3 - Coordination Diagram and the best SP sequence Fig. 4 - Initial position (1), intermediate (2) and goal position

No context found.

S. F. Smith. A Learning System Based on Genetic Adaptive Algorithms. PhD thesis, University of Pittsburgh, Pittsburgh PA, 1980.

No context found.

Smith, S., (1980), A Learning System Based on Genetic Adaptive Algorithms, PhD Dissertation, University of Pittsburgh.

No context found.

S. F. Smith. A Learning System Based on Genetic Adaptive Algorithms. PhD thesis, University of Pittsburgh, 1980.

No context found.

Smith, Steven F. A Learning System Based on Genetic Adaptive Algorithms. PhD Dissertation. Pittsburgh:

First 50 documents  Next 50

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