Srinvas, M. and L.M. Patnaik #1994#. Adaptive probabilities of crossover and mutation in genetic algorithms. IEEE Transactions on Systems, Man, and Cybernetics 24, 656# 667.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....nodes in route, D route is the distance travelled by the specified route, and D is a predefined small positive value to ensure non zero fitness scores Thus, the higher the fitness value, the more optimum the route. 3. 3 Selection Scheme We investigate three different selection schemes [17] used to selected the routes: Roulette Wheel Selection . Tournament Selection . Stochastic Remainder Each selection scheme employs a different approach. Roulette Wheel Selection is based on the random selection of generating a possible value between 0 and 1, multiplied by the sum of fitness ....

....the best performance in determining optimum route. 3.4 Crossover Operation Since chromosomes represent solutions to routes, a valid gene sequence of the chromosome (valid route) cannot contain two similar nodes. Different crossover types for permutation, order based problems are covered in [4,5,17]. Crossover techniques like PMX and Order Crossover will not only destroy the route but also result in invalid source and destination nodes. A secondary constraint is selected routes used to perform crossover operation might contain different route lengths (in terms of number of nodes required to ....

Srinivas, M. & Patnaik, L.M., "Adaptive probabilities of crossover and mutation in genetic algorithms", Tran Sys, Man Cybernetics, V24 N4, Apr94, pp 656-667.

....chains in order to determine the optimal parameter values [25, 52, 62] However, all of these studies consider only binary encoding schemes, and only small problems are used in the empirical studies. An alternate approach is to allow the operator probabilities to adapt during the GA s evolution [13, 61]. 1.3.5 Final Comments GAs GAs have gained popularity for solving optimisation problems. However, De Jong [38] emphasises that GAs are not function optimisers, but can be adapted to work as such. Davis [13, 14] and Michalewicz [48] go even further, stressing that a GA must be adapted to suit ....

....into the GCA will improve its performance, but for the same reasons will not match the performance of the heuristic on its own. There are four possible areas of improvement for the genetic clustering algorithms. Firstly, the GCAs could be adapted to control their own operator probabilities [13, 61]. This may even involve different rates according to the fitness of each chromosome, so that, for example, less fit chromosomes are more likely to undergo mutation. This would remove the need to determine good parameter values, and should enhance the performance of the GCAs. Secondly, better ....

M. Srinivas and L. M. Patnaik. Adaptive probabilities of crossover and mutation in genetic algorithms. IEEE Transactions on Systems, Man, and Cybernetics, 24(4):656--667, 1994.

....by values of Crossover Probability p c and Mutation Probability 12 pm , and the type of crossover applied [21] Increasing values of p c and pm promotes exploration at the expense of exploitation. To accomplish this trade off between exploration and exploitation in a different manner, Srinivas [22] designed an algorithm that could vary p c and pm adaptively in response to the fitness values of the solutions: p c and pm are increased when the population tends to get stuck at a local optimum and are decreased when the population is scattered in the solution space. Here we employ Srinivas s ....

M. Srinvas, L.M. Patnaik, Adaptive probabilities of crossover and mutation in Genetic Algorithms, IEEE Transaction on System, Man and Cybernetics, 24 (1994) 656-667. 17

....evaluation of a chromosome and offspring allocation. Evaluation is performed by measuring the above defined fitness value; offspring generation is then done by allocating to each chromosome, a number of offspring proportional to its fitness. GCSA implements the stochastic remainder technique [30]: a chromosome is assigned offspring according to the integer part of the proportionate fitness (f F) value in a deterministic way and the fractional parts are put in a roulette wheel 1 , for determining the 1 Roulette wheel selection allocates a sector of the wheel equal to 2pf F to every ....

Srinivas, M., Patnaik, L.M. "Adaptive Probabilities of Crossover and Mutation in Genetic Algorithms". IEEE Trans. Systems, Man and Cybernetics, vol. 24(4), 656-667, 1994.

....In the meanwhile, 5] suggest a slight advantage of Gaussian perturbations over lognormal updates when selfadaptively evolving nite state machines. Hinterding et al. 63] apply self adaptation of the mutation step size for optimizing numeric functions in a real valued GA. Srinivas and Patnaik [125] replace an individual by its child. Each chromosome has its own probabilities, pm and p c , added to their bitstring. Both are adapted in proportion to the population maximum and mean tness. B ack [9] 8] self adapts the mutation rate of a GA by adding a rate for the pm , coded in bits, to ....

....di erent crossovers, 2 point crossover and uniform crossover, by adding one extra bit to each individual (see Section IV) This extra bit decides which type of crossover is used for that individual. O spring will inherit the choice for its type of crossover from its parents. Srinivas and Patnaik [125] replace an individual by its o spring. Each chromosome has its own probabilities, pm and p c , added to their bitstring. Both are adapted in proportion to the population maximum and mean tness. In Scha er and Morishima [108] the number and locations of crossover points was selfadapted. This was ....

M. Srinivas and L.M. Patnaik. Adaptive probabilities of crossover and mutation in genetic algorithms. IEEE Transactions on Systems, Man, and Cybernetics, vol. 24(4):17{ 26, 1994.

....Their effects vary with different problems and their implementation strategies need ad hoc modifications with respect to different situations. A critical problem in studying premature convergence is the identification when it has occured and the characterization of its extent. Srinivas and Patnaik [8], for example, use the difference between the average and maximum fitness values as a yardstick to measure premature convergence, though not a measure of genetic diversity, and vary adaptively the crossover and mutation probabilities according to the measurement. On the other hand, the term ....

....ja m Gamma 1 2 j, even smaller. Remark 3.4 From Theorem 3.3 and Corollary 3.1, we can see that the probability of premature convergence at a gene position is independent of the crossover probability. So the method of adapting crossover probability to prevent premature convergence presented in [8] seems to bare no theoretical support. Adapting the crossover probability can merely speed up the search of the minimum schema containing the current population. From Theorem 3.1, we can also get the following corollary which partly answers the question of where a CGA most likely converges to. ....

M.Srinivas and L.M.Patnaik, "Adaptive probabilities of crossover and mutation in genetic algorithms", IEEE Trans. on Systems, Man and Cybernetics, vol.24,no.4,pp.656-667,1994.

....Proceedings E: Comput. Digit. Tech. 161] IEE Proceedings, Computers and Digital Techniques, 43] IEEE Transactions on Computing, 59] IEEE Transactions on Image Processing, 134] IEEE Transactions on Knowledge and Data Engineering, 87] IEEE Transactions on Systems, Man, and Cybernetics, [32] IEEE Transactions on Systems, Man, and Cybernetics A: Syst. Humans. 81] Indian J. Phys. B, 145] Inf. Process. Lett. Netherlands) 46] Inf. Sci. 102] Inf. Sci. USA) 119, 133] Information Processing Letters, 19, 163] Information Sciences, 16, 34] Int. J. High Speed Comput. ....

....G. A. Vijayalakshmi, 105] Pal, N. R. 102] Pal, Nikhil R. 16, 102, 132] Pal, S. K. 119, 133, 149] Pal, Sankar K. 22, 33, 34, 35, 44, 56, 58] Paranjothi, R. 60] Parbhane, Rupali V. 13] Parthasarathy, Guturu, 21] Patnaik, L. M. 23, 41, 87, 166, 167, 168] Patnaik, Lalit M. [25, 32, 76] Ponnambalam, S. G. 114, 128, 128] Ponnavaikko, M. 60] Prakash, M. 48, 98] Prakash, V. Syam, 42] Prasad, Rajendra, 144] Prashanth, N. 37] Pratihar, D. K. 150] Quapp, W. 12] R, Rajani, Joshi, 71] Raghavan, S. V. 155] Rajan, e.g. 37, 49] Rajarajan, N. 88] Rajasekaran, S. ....

[Article contains additional citation context not shown here]

M. Srinivas and Lalit M. Patnaik. Adaptive probabilities of crossover and mutation in genetic algorithms. IEEE Transactions on Systems, Man, and Cybernetics, 24(4):656-667, April 1994. ga94bSrinivas.

....in which chromosomes of a population become homogeneous and converge to a sub optimal chromosome. Scaling involves re adjusting the fitness values of solutions in order to sustain a steady selective pressure in the population so that a premature convergence may be avoided. Srinivas and Patnaik [175] proposed an adaptive method to tune parameters on the fly based on the idea of sustaining diversity in a population without affecting its convergence properties. Their algorithm protects the best solutions in each generation from being disrupted by crossover and mutation. Extensive experiments ....

....[181] has suggested, if the parallel processors executing a parallel genetic algorithm use heterogeneous control parameters, the diversity of the global population can be more effectively sustained. To implement this strategy, we use adaptive control parameters as suggested by Srinivas et al. [175]. The adaptive crossover rate is defined as follows: where is the maximum fitness value in the local population, is the average fitness value, is the fitness value of the fitter parent for the crossover, and is a positive real constant less than 1. The adaptive mutation rate is defined as ....

M. Srinivas and L.M. Patnaik, "Adaptive Probabilities of Crossover and Mutation in Genetic Algorithms," IEEE Transactions on Systems, Man and Cybernetics, vol. 24, no. 4, Apr. 1994, pp. 656-667.

.... Pattern Analysis and Machine Intelligence, 812, 109] IEEE Transactions on Power Delivery, 1148] IEEE Transactions on Power Systems, 737, 1020, 1058, 1063, 1070, 1094, 1124, 1139, 688] IEEE Transactions on Signal Processing, 835, 1205] IEEE Transactions on Systems, Man, and Cybernetics, [717, 493, 1102, 429, 1214] IEEE Transactions on Systems, Man, and Cybernetics Part B: Cybernetics, 1067, 1080] IEEE Transactions on Systems, Man, and Cybernetics A: Syst. Humans. 542] IEEE Transactions on Systems, Man, and Cybernetics B, Cybernetics, 1150] IEEE Transactions on Systems, Man, and Cybernetics, ....

....Park, Se Hyun, 889] Park, Seung Ho, 951] Park, Taehoon, 769] Park, Y. M. 750] Park, Young Jun, 749] Park, Young Moon, 792] Park, YoungJa, 905] Park, Young Moon, 737, 746, 771, 957] Parthasarathy, Guturu, 482] Patnaik, L. M. 484, 502, 548, 628, 629, 630] Patnaik, Lalit M. [486, 493, 537] Pei, Liu, 197, 247] Pei, Min, 315] Peichao, Zhang, 352] Peifa, Jia, 314] Peng, Tian, 212] Peng, Yi, 282] Perez, Rafael, 745] Phua, K. H. 683] Ping, Liu, 312, 339] Ping, Wang, 144] Pingfan, Yan, 133, 149] Polvichai, J. 1230, 1231] Ponnambalam, S. G. 558, 589, 589] ....

[Article contains additional citation context not shown here]

M. Srinivas and Lalit M. Patnaik. Adaptive probabilities of crossover and mutation in genetic algorithms. IEEE Transactions on Systems, Man, and Cybernetics, 24(4):656-667, April 1994. ga94bSrinivas.

....homogeneous and converge to a sub optimal chromosome. Scaling involves re adjusting the fitness values of solutions in order to sustain a steady selective pressure in the population so that a premature convergence may be avoided. To tune the control parameters on the fly, Srinivas and Patnaik [35] proposed an adaptive method which is driven by the idea of sustaining diversity in a population without affecting its convergence properties. Their algorithm [35] protects the best solutions in each generation from being disrupted by crossover and mutation. Extensive experiments have shown that ....

....in the population so that a premature convergence may be avoided. To tune the control parameters on the fly, Srinivas and Patnaik [35] proposed an adaptive method which is driven by the idea of sustaining diversity in a population without affecting its convergence properties. Their algorithm [35] protects the best solutions in each generation from being disrupted by crossover and mutation. Extensive experiments have shown that this adaptive strategy can help prevent GA s getting stuck at a local minimum. 3.4 Parallel Genetic Algorithms The inherent parallelism in GAs can be exploited to ....

[Article contains additional citation context not shown here]

M. Srinivas and L.M. Patnaik, "Adaptive Probabilities of Crossover and Mutation in Genetic Algorithms," IEEE Trans. Sys., Man and Cybernetics, vol. 24, no. 4, Apr. 1994, pp. 656-667.

.... f) if f f k 2 otherwise (6.9) where f MAX and f are the maximum and average fitness in the population, and f is the fitness of the individual under consideration. There are three disadvantages to this approach. The first is that it requires the addition of two parameters, k 1 and k 2 . Srinivas and Patnaik set both of these to 0:5 in order to completely disrupt solutions with average or below average fitness. It is not clear how a mutation rate of 0.5 completely disrupts an individual. The second disadvantage is that when the population is saturated, the mutation rate is actually undefined according to this ....

M. Srinivas and L. M. Patnaik (1994a). Adaptive probabilities of crossover and mutation in genetic algorithms. IEEE Transactions on Systems, Man and Cybernetics 4, 656-- 667.

....stages of the evolutionary process. The operators should adapt (e.g. adaptive crossover Schaffer Morishima [121] Spears [131] This is true especially for time varying fitness landscapes. ffl control parameters. There have been various experiments aimed at adaptive probabilities of operators [32, 78, 132]. However, much more remains to be done. The action of determining the variables and parameters of an EA to suit the problem has been termed adapting the algorithm to the problem, and in EAs this can be done while the algorithm is finding the problem solution. In [71] a comprehensive ....

M. Srinivas and L.M. Patnaik. Adaptive probabilities of crossover and mutation in genetic algorithms. IEEE Transactions on Systems, Man, and Cybernetics, vol. 24(4):pp. 17--26, 1994.

....paths and the decisions are made according to the value of the respective probability values p 1 ; p 2 ,p 3 . These parameters are computed at the beginning of every generation starting from parameters describing the status of the current generation; they can be regarded as adaptive parameters [17]. p2 Cut Splice 2 N Y Selection Crowding Mutation 1 Mutation Offspring Crossover Replacing Selection Simple p1 1 p1 1 p2 1 p3 p3 Deterministic Evaluated Pop. Tournament Binary Genetic Operators Comparison Operators Unary Genetic Operators Selection Operators Figure 1: A schematic ....

M. Srinivas and L. Patnaik. Adaptive Probabilities of Crossover and Mutation in Genetic Algorithm. IEEE Transactions on System, Man, and Cybernetics, 24(4), Apr 1994.

....current topology of the landscape being searched (e.g. adaptive crossover [53, 60] This is especially important for time varying fitness landscapes. Control parameters. There were already experiments aimed at these issues: adaptive population sizes [2] or adaptive probabilities of operators [10, 34, 61]. However, much more remains to be done. It seems that this is one of the most promising directions of research; after all, the power of evolutionary algorithms lies in their adaptiveness. The adaptiveness is also the key issue in the EP N project. For example, the current version of the system ....

Srinivas, M. and Patnaik, L.M., Adaptive Probabilities of Crossover and Mutation in Genetic Algorithms, IEEE Transactions on Systems, Man, and Cybernetics, Vol.24, No.4, 1994, pp.17--26.

....and we go into the parameter setting adaptation in depth. 2.1 Classifications of Adaptive Genetic Algorithms Adaptive methods may be classified following two different ways. The first one takes into account the GA issues that are adapted throughout the GA run: adaptive parameter settings ([32, 41, 7, 63, 11, 20, 14, 59, 60, 12, 3, 5, 38, 53, 1, 55, 34]) adaptive genetic operators ( 42, 30, 31, 44] adaptive genetic operator selection ( 48, 54, 52, 62, 51] adaptive representation ( 46, 61, 50] and adaptive fitness function ( 22, 42, 35, 45] The other one, suggested by Spears ( 54] concerns the interaction between the ....

....of finding the optimum solutions. To do so, pm decreases from the current value to a minimum value of 0:001 and p c is help at a constant value of 0:6. The experiments showed that dynamic GA was near an ideal behaviour defined by Li et al. with respect to the level of diversity generated. In [55], the use of adaptive probabilities of crossover and mutation was recommended to achieve the twin goals of maintaining diversity in the population and sustaining the convergence capacity of the GA. AGA was proposed, an adaptive GA where p c and pm are varied depending on the fitness values of the ....

[Article contains additional citation context not shown here]

M. Srinivas, L.M. Patnaik, Adaptive probabilities of crossover and mutation in genetic algorithms. IEEE Trans. on Systems, Man, and Cybernetics 24(4) (1994) 656-667.

No context found.

Srinvas, M. and L.M. Patnaik #1994#. Adaptive probabilities of crossover and mutation in genetic algorithms. IEEE Transactions on Systems, Man, and Cybernetics 24, 656# 667.

No context found.

M. Srinivas and L.M. Patnaik, "Adaptive Probabilities of Crossover and Mutation in Genetic Algorithms", IEEE Transaction on System, Man and Cybernetics, IEEE, 24, 1994, pp. 656-667.

No context found.

Srinivas, M., Patnaik, L.: Adaptive Probabilities of Crossover and Mutation in Genetic Algorithms . IEEE Transactions on Systems, Man, and Cybernetics 24(4) (1994) 656-667

No context found.

Srinivas, M., Patnaik, L.: Adaptive Probabilities of Crossover and Mutation in Genetic Algorithms . IEEE Transactions on Systems, Man, and Cybernetics 24(4) (1994) 656--667

No context found.

M.Srinivas and L.M.Patnaik, \Adaptive probabilities of crossover and mutation in genetic algorithms", IEEE Trans. on Systems, Man and Cybernetics, vol.24,no.4,pp.656-667,1994.

No context found.

Srinivas, M., Patnaik, L.: Adaptive Probabilities of Crossover and Mutation in Genetic Algorithms . IEEE Transactions on Systems, Man, and Cybernetics 24(4) (1994) 656--667

No context found.

Srinivas, M., Patnaik, L.: Adaptive Probabilities of Crossover and Mutation in Genetic Algorithms . IEEE Transactions on Systems, Man, and Cybernetics 24(4) (1994) 656--667

No context found.

Srinivas M. et al.: Adaptive probabilities of crossover and mutation in genetic algorithms. IEEE Transactions on Systems, Man, and Cybernetics Vol.24 No.4 (1994) 656--667

No context found.

M. Srinivas and L. Patnaik. Adaptive probabilities of crossover and mutation in genetic algorithms. IEEE Transactions on Systems, Man, and Cybernetics, 24(4):656--667, 1994. 14

No context found.

M. Srinivas and L.M. Patnaik. Adaptive probabilities of crossover and mutation in genetic algorithms. IEEE Transactions on Systems, Man, and Cybernetics, vol. 24(4):17--26, 1994.

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