Applying the Implicit Redundant Representation Genetic Algorithm in an Unstructured Problem Domain

This paper describes FOX-GA, a genetic algorithm (GA) that generates and evaluates plans in the complex domain of military maneuver planning. FOX-GA's contributions are to demonstrate an effective application of GA technology to a complex real world planning problem, and to provide an understanding of the properties needed in a GA solution to meet the challenges of decision support in complex domains. Previous obstacles to applying GA technology to maneuver planning include the lack of efficient algorithms for determining the fitn ess of plans. Detailed simulations would ideally be used to evaluate these plans, but most such simulations typically require several hours to assess a single plan. Since a GA needs to quickly generate and evaluate thousands of plans, these methods are too slow. To solve this problem we developed an efficient evaluator (wargamer) that uses course-grained representations of this problem domain to allow appropriate yet intelligent trade-offs between computational efficiency and accuracy. An additional challenge was that users needed a diverse set of significantly different plan options from which to choose. Typical GA's tend to develop a group of “best” solutions that may be very similar (or identical) to each other. This may not provide users with sufficient choice. We addressed this problem by adding a niching strategy to the selection mechanism to insure diversity in the solution set, providing users with a more satisfactory range of choices. FOX-GA's impact will be in providing decision support to time constrained and cognitively overloaded battlestaff to help them rapidly explore options, create plans, and better cope with the information demands of modern warfare.

Download Full-text

Completely Automated CNN Architecture Design Based on Blocks

10.26686/wgtn.13158305.v1 ◽

2020 ◽

Author(s):

Y Sun ◽

Bing Xue ◽

Mengjie Zhang ◽

GG Yen

Keyword(s):

Genetic Algorithm ◽

Classification Accuracy ◽

Domain Knowledge ◽

State Of The Art ◽

Classification Performance ◽

Data Sets ◽

Computational Resource ◽

Problem Domain ◽

Benchmark Data ◽

Interested User

© 2019 IEEE. The performance of convolutional neural networks (CNNs) highly relies on their architectures. In order to design a CNN with promising performance, extensive expertise in both CNNs and the investigated problem domain is required, which is not necessarily available to every interested user. To address this problem, we propose to automatically evolve CNN architectures by using a genetic algorithm (GA) based on ResNet and DenseNet blocks. The proposed algorithm is completely automatic in designing CNN architectures. In particular, neither preprocessing before it starts nor postprocessing in terms of CNNs is needed. Furthermore, the proposed algorithm does not require users with domain knowledge on CNNs, the investigated problem, or even GAs. The proposed algorithm is evaluated on the CIFAR10 and CIFAR100 benchmark data sets against 18 state-of-the-art peer competitors. Experimental results show that the proposed algorithm outperforms the state-of-the-art CNNs hand-crafted and the CNNs designed by automatic peer competitors in terms of the classification performance and achieves a competitive classification accuracy against semiautomatic peer competitors. In addition, the proposed algorithm consumes much less computational resource than most peer competitors in finding the best CNN architectures.

Download Full-text

Completely Automated CNN Architecture Design Based on Blocks

10.26686/wgtn.13158305 ◽

2020 ◽

Author(s):

Y Sun ◽

Bing Xue ◽

Mengjie Zhang ◽

GG Yen

Keyword(s):

Genetic Algorithm ◽

Classification Accuracy ◽

Domain Knowledge ◽

State Of The Art ◽

Classification Performance ◽

Data Sets ◽

Computational Resource ◽

Problem Domain ◽

Benchmark Data ◽

Interested User

© 2019 IEEE. The performance of convolutional neural networks (CNNs) highly relies on their architectures. In order to design a CNN with promising performance, extensive expertise in both CNNs and the investigated problem domain is required, which is not necessarily available to every interested user. To address this problem, we propose to automatically evolve CNN architectures by using a genetic algorithm (GA) based on ResNet and DenseNet blocks. The proposed algorithm is completely automatic in designing CNN architectures. In particular, neither preprocessing before it starts nor postprocessing in terms of CNNs is needed. Furthermore, the proposed algorithm does not require users with domain knowledge on CNNs, the investigated problem, or even GAs. The proposed algorithm is evaluated on the CIFAR10 and CIFAR100 benchmark data sets against 18 state-of-the-art peer competitors. Experimental results show that the proposed algorithm outperforms the state-of-the-art CNNs hand-crafted and the CNNs designed by automatic peer competitors in terms of the classification performance and achieves a competitive classification accuracy against semiautomatic peer competitors. In addition, the proposed algorithm consumes much less computational resource than most peer competitors in finding the best CNN architectures.

Download Full-text

Implicit Representation in Genetic Algorithms Using Redundancy

Evolutionary Computation ◽

10.1162/evco.1997.5.3.277 ◽

1997 ◽

Vol 5 (3) ◽

pp. 277-302 ◽

Cited By ~ 16

Author(s):

Anne M. Raich ◽

Jamshid Ghaboussi

Keyword(s):

Genetic Algorithm ◽

Genetic Algorithms ◽

Fitness Function ◽

Search Space ◽

Function Evaluation ◽

Diverse Population ◽

Implicit Representation ◽

Essential Information ◽

Redundant Representation ◽

Better Than

A new representation combining redundancy and implicit fitness constraints is introduced that performs better than a simple genetic algorithm (GA) and a structured GA in experiments. The implicit redundant representation (IRR) consists of a string that is over-specified, allowing for sections of the string to remain inactive during function evaluation. The representation does not require the user to prespecify the number of parameters to evaluate or the location of these parameters within the string. This information is obtained implicitly by the fitness function during the GA operations. The good performance of the IRR can be attributed to several factors: less disruption of existing fit members due to the increased probability of crossovers and mutation affecting only redundant material; discovery of fit members through the conversion of redundant material into essential information; and the ability to enlarge or reduce the search space dynamically by varying the number of variables evaluated by the fitness function. The IRR GA provides a more biologically parallel representation that maintains a diverse population throughout the evolution process. In addition, the IRR provides the necessary flexibility to represent unstructured problem domains that do not have the explicit constraints required by fixed representations.

Download Full-text