Holland H. J., Adaptation in Natural and Artificial Systems, an introductory analysis with application to biology, control and artificial intelligence. Ann Arbor, The university of Michigan Press, 1975.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....the computers are increasing this disadvantage is diminishing. Furthermore, most non gradient methods are well suited for implementation on parallel processors. There is a large number of non derivative methods. For example, the Complex method developed by Box [8] in the 60 s, genetic algorithms [35] or the similar evolutionary algorithms [68] both developed in the early 70 s by Holland and Rechenberg respectively. Simulated annealing was then developed by Kirkpatrick [44] in the early 80 s. Methods that are more recent include Tabu search, developed by Glover [27] in 1989, which have been ....

....the Complex method could also be applied to mixed continuous and discrete variable problems. 5. 2 Genetic algorithms Genetic algorithms (GAs) and the closely related evolutionary algorithms are a class of non gradient methods which has grown in popularity ever since Rechenberg [68] and Holland [35] first published their work on the subject in the early 70 s. For a more comprehensive study of genetic algorithms, see Goldberg s [28] splendid book on the subject. Genetic algorithms are modeled after mechanisms of natural selection. Each optimization real number or a string of bits. The ....

HOLLAND H. J., Adaptation in Natural and Artificial Systems, an introductory analysis with application to biology, control and artificial intelligence, The university of Michigan Press, Ann Arbor, USA, 1975.

....must deal with hard tasks, often characterized by an abundance of local optima, and therefore they depend on efficient diversification strategies. For example stochastic algorithms like the Simulated Annealing (SA) of [23] place non zero probabilities for upward moving steps, Genetic Algorithms [22] use recombination and mutation mechanisms inspired by the natural evolution, deterministic algorithms like the Tabu Search (TS) of [18, 19] prohibit the repetition of previously visited configurations in an explicit manner. Now, a possible approach to solve a hard continuous optimization task ....

J. H. Holland, Adaptation in natural and artificial systems, an introductory analysis with applications to biology, control, artificial intelligence (University of Michigan Press, Ann Arbor, 1975).

....and control dimensions lead to as many different search algorithms. For instance, the class of algorithms using variation methods inspired by reproduction in nature and Darwinianlike selection is called Evolutionary Computation (EC) of which GP is an instance. Genetic algorithms (GAs) [Holland, 1975], evolution strategies (ES) Schwefel, 1981; Back and Schwefel, 1993] and evolutionary programming (EP) Fogel et al. 1966; Fogel, 1995] are special instances of the EC class differentiated by the use of linear bit string representations (the first) or more complex linear representations and ....

....evolutionary processes in order to build artificial systems has considerably grown and has outlined alternative solutions to hard search and optimization problems. Several distinct approaches currently coexist under the name of evolutionary algorithms (EA) They are: genetic algorithms (GA) [Holland, 1975; Holland, 1992] evolutionary programming (EP) Fogel et al. 1966; Fogel, 1995] evolution strategies (ES) Schwefel, 1981; Back and Schwefel, 1993] genetic programming (GP) Koza, 1989; Koza, 1992] and evolutionary reinforcement learning (or classifier systems) ERL) Goldberg, 1989; Wilson, ....

[Article contains additional citation context not shown here]

John H. Holland, Adaptation in Natural and Artificial Systems, An Introductory Analysis with Applications to Biology, Control and Artificial Intelligence, The University of Michigan, 1st edition, 1975.

....a chromosome is a vector of the form hr 1 ; r 2 ; r m i, where r 1 ; r 2 ; r m are real numbers whose lower and upper bounds are defined by the user. Despite the well known fact that the binary alphabet offers the maximum number of schemata per bit of information of any coding [28], the implicit parallelism property of genetic algorithms does not preclude the use of alphabets of higher cardinality. Whereas theoreticians claim that small alphabets should be more effective than large alphabets, practitioners have shown through a considerable amount of real world ....

John H. Holland. Adaptation in Natural and Artificial Systems. An Introductory Analysis with Applications to Biology, Control and Artificial Intelligence. MIT Press, Cambridge, Massachusetts, 1992.

....logic circuits, which contain no memory elements and no feedback paths. However, the approach proposed is general enough as to allow its generalization to other (more complex) circuits. 2 Previous Work A general search technique inspired by natural evolution, called the genetic algorithm (GA) [15], has been widely used for optimization tasks [9] and is known to be a very powerful tool in certain domains. In our current work we wish to find a way to use the GA as a design tool, with particular emphasis in the design of combinational circuits. The design process for combinational logic ....

....in the late 1960s, while working at the University of Michigan, he developed a technique that allowed computer programs to mimic the process of evolution. Originally, this technique was called reproductive plans, but the term genetic algorithm became popular after the publication of his book [14] [15]. More information on genetic algorithms may be found in the books by Goldberg [9] Michalewicz [27] and Mitchell [31] A genetic algorithm for a particular problem must have the following five components [27] 7 Input 1 Input 2 Gate Type Figure 2: Encoding used for each of the matrix ....

John H. Holland. Adaptation in Natural and Artificial Systems. An Introductory Analysis with Applications to Biology, Control and Artificial Intelligence. MIT Press, Cambridge, Massachusetts, 1992.

.... 32 34, 36, 39, 40] In the past these problems have been solved using simulated annealing techniques [6, 7] dual methods [35] and branch bound methods using sequential quadratic programming [21, 22, 34] The genetic algorithm is a rather recent means with which to solve optimal design problems [5, 15, 17, 18, 30, 31] and it is based on Darwin s Theory of Evolution. In the classical genetic algorithm formulation, possible design Voss Foley: 2 configurations are termed individuals and their characteristics are defined using genetic coding (usually binary strings representing design variables) Each ....

....a priori problem specific information being required. However, this generality can cause the algorithm to spend time searching regions of the solution space that are known to be unprofitable. A question then arises: Given two designs for the same structure, what are the structure s building blocks [14, 18 20] and in what meaningful ways could they be exchanged in the crossover operations typically found in genetic algorithms Traditional multistory buildings tend to have their heaviest members near the base and their lightest members near the top. They also tend to gradually change the weight of ....

[Article contains additional citation context not shown here]

Holland, J. H., Adaptation In Natural And Artificial Systems, (An Introductory Analysis With Applications To Biology, Control, and Artificial Intelligence). Cambridge, London: MIT Press, 1975.

....logic circuits, which contain no memory elements and no feedback paths. However, the approach proposed is general enough as to allow its generalization to other (more complex) circuits. 2 Previous Work A general search technique inspired by natural evolution, called the genetic algorithm (GA) [12], has been widely used for optimization tasks [9] and is known to be a very powerful tool in certain domains. In our current work we wish to find a way to use the GA as a design tool, with particular emphasis in the design of combinational circuits. The design process for combinational logic ....

....in the late 1960s, while working at the University of Michigan, he developed a technique that allowed computer programs to mimic the process of evolution. Originally, this technique was called reproductive plans, but the term genetic algorithm became popular after the publication of his book [11] [12]. More information on genetic algorithms may be found in the books by Goldberg [9] Michalewicz [23] and Mitchell [25] A genetic algorithm for a particular problem must have the following five components [23] 1. A representation for potential solutions to the problem. Input 1 Input 2 Gate Type ....

John H. Holland. Adaptation in Natural and Artificial Systems. An Introductory Analysis with Applications to Biology, Control and Artificial Intelligence. MIT Press, Cambridge, Massachusetts, 1992.

....the minimum description length principle to analyze the power of approaches based on automatic discovery of functions. We conclude by analyzing an example, mentioning related work and considering directions for future work. 2 BUILDING BLOCKS IN GP In Genetic Algorithms (GA) the schemata theorem [Holland, 1992] , Goldberg, 1989] summarizes the effect of fitness proportionate reproduction, crossover and mutation. Schemata are template strings representing sets of individuals in the search space. Schemata are defined by strings over the (usually binary) alphabet extended with a don t care symbol. The ....

John H. Holland. Adaptation in Natural and Artificial Systems, An Introductory Analysis with Applications to Biology, Control and Artificial Intelligence. MIT Press, 1992.

....it should explore actions at random, in order to experiment with more state action pairs. In this case the system acts to gain information or experience, that is explores the search space. The choice of what to do next exemplifies a well known problem, the exploration exploitation trade off (see [5]) Our approach is to use ACK to describe explicit exploratory behaviors. An example of such a goal is map building. A better estimation of surrounding object positions can affect the execution of system knowledge sources, resulting in improved overall behavior. Map building is one of the ....

John H. Holland. Adaptation in Natural and Artificial Systems, An Introductory Analysis with Applications to Biology, Control and Artificial Intelligence. MIT Press, second edition, 1992.

....mathematical essence of life as it evolves from physics and biochemistry. He succeeded to make the first step, showing that the exact reproduction of universal Turing machines is possible in a particular deterministic model Universe. These results are consistent with Holland s model of Universe [23, 24] and his claims concerning a possibility to demonstrate that self replicating systems can emerge from unorganized initial states . 8 Following this path of thought it may be possible to formulate a way to 7 Teleology is the idea that physical processes can be determined by, or drawn towards, ....

....very similar to the well known example on movies: the film appears to the human eye to be continuous, but in reality it is discrete. 12 The reason of this 9 Recall that # and K encode the same amount of information, but di#er in the structuralization of this information 10 See also Holland [23, 24]. 11 For instance, experiments in physics cannot measure intervals of time less than 10 26 seconds. 12 It advances one frame at a time. apparent paradox comes from the fact that human eyes cannot resolve the short time intervals between frames. For a detailed discussion we quote Mellor ....

J. H. Holland. Adaptation in Natural and Artificial Systems. An Introductory Analysis with Applications to Biology, Control, and Artificial Intelligence, The University of Michigan Press, Ann Arbor, 1975.

....additional method of providing feedback to the filtering system. The user is no longer constrained to providing feedback only to the documents retrieved by the profiles, but can also pro actively train the system using documents she found. ffl This thesis validates the use of Genetic Algorithms [18] for modeling adaptive and exploratory behavior in Filtering systems. ffl Experimental results show that using only relevance feedback is sufficient for specializing to user interests, but not satisfactory for modeling adaptive behavior. 1.3 Overview of this document The rest of this thesis is ....

....are matched with the contents of the documents i.e. the agents use cognitive filtering. The algorithm used by the agent is described in the following chapter. The learning mechanism used in the information filtering agents is motivated by research in Genetic Algorithms and Artificial Evolution [1, 6, 15, 18, 19]. IF is effectively a dynamically changing search problem. Searching a large and changing space involves a trade off between two objectives: i) exploiting the currently available solution and (ii) further exploring the search space for a possibly better solution. Hill Climbing is an example of a ....

[Article contains additional citation context not shown here]

Holland, J.H., Adaptation in Natural and Artificial Systems, An Introductory Analysis with Applications to Biology, Control and Artificial Intelligence, University of Michigan Press, Ann Arbor, 1975.

....an agent in a dynamic and non deterministic environment. The automatic discovery of subroutines can help scale up the GP technique to complex problems. 1 Introduction Holland hypothesized that genetic algorithms (GAs) achieve their search capabilities by means of block processing (see [Holland, 1975], Goldberg, 1989] Blocks are relevant pieces of a solution that can be assembled together, through crossover, in order to generate problem solutions. Goldberg argues in favor of the hypothesis of building block processing by looking also for arguments in nature: simple life forms gave way ....

John H. Holland, Adaptation in Natural and Artificial Systems, An Introductory Analysis with Applications to Biology, Control and Artificial Intelligence, The University of Michigan, 1st edition, 1975.

....expressed in this material are those of the authors and do not necessarily reflect the views of these agencies. 1 Introduction Complexity of evolved structures is a non issue in most of the recent evolutionary computation (EC) literature. EC techniques such as genetic algorithms (GAs) [Holland, 1975], evolutionary programming (EP) Fogel et al. 1966; Fogel, 1995] and evolution strategies (ES) Back et al. 1991] use mostly fixed length encodings of the structures to be evolved. This design decision seriously limits their applicability to the domain of parametric problems. Many applications ....

....an optimal string combines exploitation (preservation of schemata) and exploration (creation of new schemata) in close to an optimal proportion. The argument relied on the analogy between the allocation of samples to schemata in the GA with the allocation of effort in the Two Armed Bandit problem [Holland, 1975; Holland, 1992] Schemata theory has been criticized for not reflecting the processing done by a GA and not being really informative. One such critique is that GA allocates trials to schemata very differently from the optimal allocation given by the Two Armed Bandit solution. This was shown on ....

John H. Holland, Adaptation in Natural and Artificial Systems, An Introductory Analysis with Applications to Biology, Control and Artificial Intelligence, The University of Michigan, 1st edition, 1975.

....survey that can be used (to a certain extent) as a self contained document for anyone interested in this area who has a previous (at least basic) knowledge of genetic algorithms in general. Those who may need additional information about genetic algorithms should refer to Goldberg [27] Holland [35], Michalewicz [54] and Mitchell [56] for more information. 2 Statement of the Problem Multiobjective optimization (also called multicriteria optimization, multiperformance or vector optimization) can be defined as the problem of finding [65] a vector of decision variables which satisfies ....

John H. Holland. Adaptation in Natural and Artificial Systems. An Introductory Analysis with Applications to Biology, Control and Artificial Intelligence. MIT Press, Cambridge, Massachusetts, second edition, 1992.

....size that solve the problem. However, by using directly the size constraint the GP algorithm is prevented from finding solutions. The algorithm improves convergence to a better optimum while maintaining speed. Exploration and exploitation are recurring themes in search and learning problems [Holland, 1992], Kaelbling, 1993] Exploitation takes place when search proceeds based on the action prescribed by the current system knowledge. Exploration is usually based on random actions, taken in order to experiment with more situations. For example, in learning classifier systems, roulette wheel action ....

....an HGP approach enables the manipulation of a population of higher diversity programs, which positively affects the efficiency of an HGP algorithm for complex problems. 4 HGP Evolution Dynamics GP evolution dynamics has been very difficult to analyze. The traditional analysis of GAs by Holland [Holland, 1992] focuses on the propagation of schemata from one generation to the next. The building block hypothesis ( Holland, 1992] Goldberg, 1989] outlines the importance of small schemata, called building blocks, in the proper functioning of a GA. More recently, crossover has been considered the ....

[Article contains additional citation context not shown here]

John H. Holland, Adaptation in Natural and Artificial Systems, An Introductory Analysis with Applications to Biology, Control and Artificial Intelligence, MIT Press, 2nd edition, 1992.

....importantly we will give a clear interpretation of what good fragments of genetic material are. GP can take advantage of its own evolutionary trace in order to discover good genetic material and use it to adapt the search process. 4. 1 Foundations In Genetic Algorithms (GA) the schemata theorem [Holland, 1992], Goldberg, 1989] summarizes the effect of fitness proportionate reproduction, crossover and mutation. Schemata are template strings representing sets of individuals in the search space. Schemata are defined by strings over the (usually binary) alphabet extended with a don t care symbol. The ....

John H. Holland, Adaptation in Natural and Artificial Systems, An Introductory Analysis with Applications to Biology, Control and Artificial Intelligence, MIT Press, second edition, 1992.

.... Currently, soft computing has not been incorporated into agent based design on a large scale [57 ] The approach taken here is to embed soft computing into agent based modeling through the concept of constrained generating procedures [8] 4 COMPLEX ADAPTIVE SYSTEMS Complex adaptive systems (CAS) [9, 10] can be defined as those systems that are composed of multiple: agents, building blocks and internal models; with the inherent capacity to exhibit perpetual novelty when subjected to a changing environment. CAS theory has been used primarily to model those systems that have been too complicated to ....

J. H. Holland, Adaptation In Natural And Artificial Systems, An Introductory Analysis With Applications To Biology, Control, and Artificial Intelligence (Cambridge, London: MIT Press, 1975).

....the process of document rating. It relies on the output of an evolutionary algorithm that represents the current interest profile of the user. Evolutionary algorithms provide methods for simulating biological evolution on a computer. Among others they include as subdisciplines genetic algorithms [22], evolution strategies [41] genetic programming [26] and artificial life [38] All have in common that they are based on the biological principle of Darwin s natural selection and survival of the fittest in that they use computational models of evolutionary processes as key elements in the ....

J. H. Holland. Adaptation in Natural and Artificial Systems. An Introductory Analysis with Applications to Biology, Control and Artificial Intelligence. MIT Press, Cambridge, Mass., 1992.

No context found.

Holland H. J., Adaptation in Natural and Artificial Systems, an introductory analysis with application to biology, control and artificial intelligence. Ann Arbor, The university of Michigan Press, 1975.

No context found.

J. H. Holland. Adaptation in Natural and Artificial Systems, An Introductory Analysis with Applications to Biology, Control and Artificial Intelligence. The University of Michigan, 1st edition, 1975.

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