System for integrating multiple genetic algorithm applications

• US 5,581,657 A
• Filed: 07/29/1994
• Issued: 12/03/1996
• Est. Priority Date: 07/29/1994
• Status: Expired due to Fees

• Current Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Original Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Inventors: Lyon, Bruce C.
• Primary Examiner(s): Downs, Robert W.
• Assistant Examiner(s): Shah, Sanjiv
• Application Number:
  US08/282,448
• Time in Patent Office: 858 Days
• Field of Search: 395/13, 364/468, 364/496
• US Class Current: 706/13
• CPC Class Codes: G06N 3/126 Evolutionary algorithms, e.g. genetic algorithms or genetic programming

Abstract

A genetic algorithm-based system for selecting a solution. Means are provided for evaluating solutions according to a preselected evaluation function. A first computer generates a first population of solutions, the first population of solutions being of a first representation scheme. A second computer generates a second population of solutions, the second population of solutions being of a second representation scheme. A translator translates solutions of the second representation scheme into equivalent solutions of the first representation scheme. Means are provided for introducing solutions from the second population into the first population of solutions.

Claims

1. A system for selecting a solution, comprising:

a first computer, for generating a first population of solutions, the first population of solutions being of a first representation scheme;

a second computer, for generating a second population of solutions, the second population of solutions being of a second representation scheme;

a translator for translating solutions of the second representation scheme into equivalent solutions of the first representation scheme;

means for introducing solutions from the second population of solutions into the first population of solutions; and

means for evaluating solutions of the first population of solutions according to a preselected evaluation function.

2. The system of claim 1, wherein the first computer is capable of generating solutions independently of the second computer.

3. The system of claim 1, wherein the first computer generates new solutions by applying a genetic algorithm technique to an initial population of solutions and selecting a subset of the new solutions and the initial solutions according to the evaluation function, the subset becoming an initial population of solutions for a subsequent application of the genetic algorithm technique.

4. The system of claim 1, further comprising an optimizing computer for generating new solutions by applying an optimization technique to a solution in a population of solutions and introducing said new solutions to the evaluation means.

5. The system of claim 4, wherein the optimizing computer replaces the solution in the population with an optimized solution according to the optimization technique if the optimized solution results in a better evaluation than the solution according to the evaluation function.

6. The system of claim 4, further comprising means for introducing optimized solutions from the optimizing computer into the first population of solutions.

7. The system of claim 1, further comprising a buffer for retaining a set of highest-evaluated solutions from the first population of solutions.

8. The system of claim 7, further comprising means for introducing solutions from the set of highest-evaluated solutions into the first population of solutions.

9. The system of claim 8, further comprising means for introducing a plurality of ,replicated solutions from the set of highest-evaluated solutions into the first population of solutions.

10. The system of claim 7, further comprising

an optimizing computer for generating new solutions by applying an optimization technique to a solution in a population of solutions;

and

means for introducing solutions from the set of highest-evaluated solutions into the optimizing computer.