| N. Macias, "Ring Around the PIG: A Parallel GA with Only Local Interactions Coupled with a Self-Reconfigurable Hardware Platform to Implement an O(1) Evolutionary Cycle for EHW," to appear in The 1999 Congress on Evolutionary Computation. |
....two methods termed Mixtrinsic evolution has been introduced at JPL [SZK00] where part of the population is evaluated extrinsically and part intrinsically. A further method exists where both the evolving design and the evolution process itself are implemented in hardware e.g. CHE [TH99] and PIG [Mac99b] When individuals of the evolution process are tested out in the actual hardware, this provides opportunities for the evolution process to explore the underlying technology s physical properties [THH96] If a design which meets specified goals is created and the evolution process is designed ....
....genotype means that each individual in the population is smaller and, as such, less storage is required to store the population. 4.2 Population and Genetic Operations An on chip implementation of the genetic algorithm involves storage of the population on chip. If, as in the PIG proposal [Mac99b] and in [PRH00] all individuals of the population are implemented in parallel on the chip then size is a limiting factor for the population. If the population is limited, this reduces the amount of genetic information available to help the evolution process towards an acceptable fitness level ....
N. Macias. Ring Around the PIG: A Parallel GA with Only Local Interactions Coupled with a Self-Reconfigurable Hardware Platform to Implement an O(1) Evolutionary Cycle for Evolvable Hardware. In Proc. of the 1999 Congress on Evolutionary Computation (CEC99), pp 1067-- 1078. IEEE, 1999.
....is independent of the population size. Execution of the RGA on a PIG was simulated for the evolution of a parity generator, 4 1 multiplexer and 3 bit counter. In each case, the algorithm was able to evolve a perfect circuit. Full details of the RGA and the experimental setup are available in [5]. While the RGA can not be implemented on a standard FPGA or other externally controlled reconfigurable device (internal control is required to keep the mating time independent of population size) it uses only one of the building blocks described above (remote cell replication) Other possible ....
N. Macias, "Ring Around the PIG: A Parallel GA with Only Local Interactions Coupled with a Self-Reconfigurable Hardware Platform to Implement an O(1) Evolutionary Cycle for EHW," to appear in The 1999 Congress on Evolutionary Computation.
....O run time in finding an encryption key. Another important feature of the Cell Matrix architecture is that a cell s internal memory can be reconfigured by other cells in the architecture. This leads to dynamic, self modifying circuits. Some interesting examples of this feature include a ringed GA [6, 7], evolvable hardware, and run time, fault tolerant hardware. The dynamic algorithm for DNA sequence alignment can also be efficiently implemented on a Cell Matrix architecture due to its support for distributed and parallel computing. The algorithm has a space complexity (number of processors) of ....
Macias, N. Ring around the PIG: a parallel GA with only local interactions coupled with self-reconfigurable hardware platform to implement an O(1) evolutionary cycle for EHW. The 1999 Congress on Evolutionary Computing (1999), 1067-1075
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