J. M. Smith, "When learning guides evolution," Nature, vol. 329, no. 6142, pp. 761--762, Oct. 1987.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....evolution is one of the most fundamental forms of adaptation, it is not surprising that the evolution of learning rules has been introduced into ANN s in order to learn their learning rules. The relationship between evolution and learning is extremely complex. Various models have been proposed [257] [271] but most of them deal with the issue of how learning can guide evolution [257] 260] and the relationship between the evolution of architectures and that of connection weights [261] 263] Research into the evolution of learning rules is still in its early stages [264] 267] 269] ....

....that the evolution of learning rules has been introduced into ANN s in order to learn their learning rules. The relationship between evolution and learning is extremely complex. Various models have been proposed [257] 271] but most of them deal with the issue of how learning can guide evolution [257] [260] and the relationship between the evolution of architectures and that of connection weights [261] 263] Research into the evolution of learning rules is still in its early stages [264] 267] 269] 270] This research is important not only in providing an automatic way of optimizing ....

J. M. Smith, "When learning guides evolution," Nature, vol. 329, no. 6142, pp. 761--762, Oct. 1987.

.... simulations of Hinton and Nowlan [10] This paper inspired much interest in the relationship between learning and evolution (at least within the evolutionary computation community genetic assimilation has not always been accepted in biology due to suspicions that such phenomena were Lamarckian [19]) Subsequent works have involved the addition of cultural learning [3] the evolution of neural network learning rates [4] and the evolution of associative learning [20] to name a few. However, understanding the original focus of Hinton and Nowlan s simulation the Baldwin e#ect is still a ....

Smith, J. M. (1987). When learning guides evolution. Nature, 329:761--762.

....In fact, I still thought of it in terms of production systems and genetic algorithms. Cellular automata were not even considered. And the stance of looking at evolutionary processes in an intertwined fashion with learning (as in [Harley 1981] Draper 1987] Hinton and Nowlan 1987] and [Smith 1987]) was still very much present. Also, the way I had approached the issue put too much emphasis on biological concepts, as if I was going to model some aspect of biological reality. Finally, the research proposal also had elements of a view of evolution from a developmental psychology point of view, ....

John Maynard Smith. When learning guides evolution. Nature, 329:761-- 762, 1987.

....It has become apparent that genetic connectionism is an ideal modeling tool in this endeavor: genetic algorithms providing an elegant model of evolution, and connectionism providing simple but powerful learning mechanisms. Belew, McInerney and Schraudolph (1990) following up suggestions by Maynard Smith (1987) and Hinton and Nowlan (1987) have demonstrated the complementary nature of the two phenomena: the presence of learning makes evolution much easier (all evolution has to do is find an appropriate initial state of a system, from which learning can do the rest) and evolutionary methods can ....

J. Maynard Smith (1987). When learning guides evolution. Nature, 329: 761--762.

.... with Drosphila subobscura which directly demonstrated modification of genomes via artificial environmental changes (see Jablonka and Lamb, 1995:31f for a detailed and accessible description of these experiments) 10 Evolutionary biologists accept the possibility of genetic assimilation (e.g. Maynard Smith, 1987, 1993:319f; Rose, 1997:217f) however, some (e.g. Dawkins, 1982:284) regard it as a hypothetical mechanism because, though it has been demonstrated experimentally, it has not been conclusively shown to occur naturally. It is extremely difficult to prove a case of natural, adaptive genetic ....

Maynard Smith, John (1987) `When learning guides evolution', Nature, vol.329, 762.

....appeared. According to Maynard Smith, the Baldwin Effect has not always been well received by biologists, partly because they have suspected it of being lamarckist [ and partly because it was not obvious that it would work. What Hinton and Nowlan have done is to answer these objections , (Maynard Smith, 1987, pp. 761 762) In other words, they have discov 2 If the conclusions prompted by an armchair thought experiment are unclear or tendentious if different thought experimenters reach different conclusions from the same premises then it is just not a very good thought experiment. Thought ....

Maynard Smith, J. (1987). When learning guides evolution. Nature, 329, 761--762.

....evolution is one of the most fundamental forms of adaptation, it is not surprising that the evolution of learning rules has been introduced into ANNs in order to learn their learning rules. The relationship between evolution and learning is extremely complex. Various models have been proposed [257, 258, 259, 260, 261, 262, 263, 264, 265, 266, 267, 268, 269, 270, 271] but most of them deal with the issue of how learning can guide evolution [257, 258, 259, 260] and the relationship between the evolution of architectures and that of connection weights [261, 262, 263] Research into the evolution of learning rules is still in its early stages [264, 265, 266, 267, ....

....ANNs in order to learn their learning rules. The relationship between evolution and learning is extremely complex. Various models have been proposed [257, 258, 259, 260, 261, 262, 263, 264, 265, 266, 267, 268, 269, 270, 271] but most of them deal with the issue of how learning can guide evolution [257, 258, 259, 260] and the relationship between the evolution of architectures and that of connection weights [261, 262, 263] Research into the evolution of learning rules is still in its early stages [264, 265, 266, 267, 269, 270] This research is important not only in providing an automatic way of optimizing ....

J. Maynard Smith, "When learning guides evolution," Nature, vol. 329, pp. 761--762, October 1987.

.... different levels of adaptation processes: one taking place during the life time of an organism and the other taking place during the evolutionary history of population of organisms [3] 18] There are quite a number of research concentrating on the relationship between evolution and learning [2] [16] [11] 19] 1] 17] The nature of interaction between the two has been shown to be complementary : the presence of learning can facilitate the process of evolution and evolutionary methods can significantly speed up the learning process. However, in these studies, the methods of learning have ....

J. Maynard Smith. When learning guides evolution. Nature, 329:761--762, 1987.

....larger multi layer network design is evolved and in [1] interaction between learning as the adaptation of individual, and evolution as the adaptation of population is observed. In fact, there are quite a number of researchers concentrating on the relationship between evolution and learning [4] [30] [22] 38] 2] 32] 3] 11] 39] 42] In these works individual network structures (representing the learning) are evolved to optimise the adaptive behavior (hence, the structure) In [33] and [34] the generation of regular networks is presented by using a recursive algorithm. However, the ....

J. Maynard Smith. When learning guides evolution. Nature, 329:761--762, 1987.

....as if it were searching for representations of the environment that best make use of the learning potential of the system. One can also think of the Baldwin Effect, where the course of evolutionary change is determined by the ability of the individuals to acquire a fit behaviour by learning (see [ 10 ] 3 ] A system like GCI shows that there is a tradeoff between the generality of the operators for construction of new attributes in CI and the complexity of the search for new attributes. With universal operators, the genetic search of the system is equally (softly) biased to all possible ....

J. Maynard-Smith. When learning guides evolution. Nature, 329:761--762, 1987.

....mean performance even in a stationary environment. One might have expected that learning would be a drawback in that case, since each generation would waste time relearning the same action function. Actually, learning can serve as a scout to help evolution find good directions in which to evolve [10, 15, 5, 7]. Such Baldwin effects may account for evolutionary reinforcement learning s edge in the stationary environment. 3 Analysis At least in this abstract and simplified case, we have seen that evolution combined with learning can outperform evolution alone across a wide range of environmental ....

....to refer to just about any sort of intra individual lifetime adaptive process. In these models, according to various criteria, learning provides an improvement compared to evolution alone. In Hinton Nowlan s model [10] learning guides an evolutionary search toward a well hidden maximum. Smith [15] points out that learning can also help compensate for the disruptive effects of sexual recombination. Fontanari Meir [7] provide simulations and analytic results of a model that shows how learning allows an evolutionary search to function successfully under much higher mutation rates. In Miller ....

Smith, J.M. (1987). When learning guides evolution. Nature, 32, 761-- 762.

.... with Drosphila subobscura which directly demonstrated modification of genomes via artificial environmental changes (see Jablonka and Lamb, 1995:31f for a detailed and accessible description of these experiments) 6 Evolutionary biologists accept the possibility of genetic assimilation (e.g. Maynard Smith, 1987, 1993:319f; Rose, 1997:217f) however, some (e.g. Dawkins, 1982:284) regard it as a hypothetical mechanism because, though it has been demonstrated experimentally, it has not been conclusively shown to occur naturally. It is extremely difficult to prove a case of natural, adaptive genetic ....

Maynard-Smith, J. (1987) `When learning guides evolution', Nature, vol.329, 762.

....Fah89, Hin87, Kee86, Kru89, MoS89, RHW86, Rum88, WiL88] 2.3. Learning Interactions The idea that learning by individuals and learning by populations interact has been around for some time [Bal96, Wad42] This idea has recently resurfaced and is beginning to be explored in the GA ANN communities [Bel89, HiN86, Smi87]. 2.4. Using GAs to Train ANNs Researchers have demonstrated the feasibility of using GAs to set the weights of feed forward networks [MoD89, Whi88, Whi89, WSB89, WhH89a, WhH89b, WhS89, Whi89] Generally the computational cost of using GA s to set the weights has contrasted well with that of ....

J. M. Smith (29 October 1987). "When learning guides evolution," Nature, vol. 329, pp. 761-762.

....as discussed above (Section 2.2.2) Notice, however, that it is the genotype that is used for breeding, so there is no genetic mechanism for passing on acquired behavior. On the other hand, acquired behavior can indirectly affect the genotype of the offspring, the so called Baldwin Effect [1, 7, 10]. The noninheritability of acquired behavior leads to an important difference between our breeding algorithm and that common in genetic algorithms. The latter typically implement breeding in distinct generations, with all of the individuals of the population being replaced at one time. A ....

Maynard-Smith, John, When Learning Guides Evolution, Nature 329 (Oct. 29 1987), 761-762.

....grammars. Both accounts of the evolution of language do not leave much room for learning. The proponents of neo Darwinian evolution do, however, leave a little elbow room for learning inasmuch as evolution through natural selection being essentially a hillclimbing process (Hinton Nowlan, 1987; Maynard Smith, 1987; Pinker Bloom, 1990) can be construed as a kind of (non Larmarckian) learning process in which a particular species searches the evolutionary problem space for good genotypes (albeit that the species has no memory of its previous searchpath since individuals with poorly adapted genotypes ....

....learning mechanisms subserving early language would then have been shaped by natural selection into our presentday language ability via natural selection. On this view, what originally had to be learned gradually became innate following the Baldwin effect (Baldwin, 1896; Hinton Nowlan, 1987; Maynard Smith, 1987), thus creating the UG that Pinker Bloom take to underlie human language. The basic idea behind the Baldwin effect is that when a species is faced with the learning of an adaptive task, certain initial settings of the learning mechanism are better than others. Individuals endowed with good ....

[Article contains additional citation context not shown here]

Maynard-Smith, J. (1987) When Learning Guides Evolution. Nature, 329, 761-- 762.

....and in particular the Baldwin effect. 3.1 The Baldwin effect Learning during one s lifetime does not directly affect one s genetic makeup; consequently, things learned during an individual s lifetime cannot be transmitted directly to its offspring. However, some evolutionary biologists (e.g. [98]) have discussed an indirect effect of learning on evolution, inspired by ideas about evolution due to Baldwin [10] The idea behind the so called Baldwin effect is that if learning helps survival, then organisms best able to learn will have the most offspring and increase the frequency of the ....

J. Maynard Smith. When learning guides evolution. Nature, 329, 1987.

No context found.

J. Maynard Smith. When Learning Guides Evolution. Nature, 329:761-762.

No context found.

J. Maynard Smith. When Learning Guides Evolution. Nature, 329:761--762, 1987.

No context found.

Maynard Smith, J. (1987). When learning guides evolution. Nature, 329, 761762.

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