Fitness 函数必须 return 一个标量

Question

我在 Matlab 上遇到一个错误，说我的适应度函数（遗传算法）应该 return 一个标量。目标是计算最佳系数 x(1) x(2) 和 x(3)：

function score= Fitness_fct(x,L,C,COV,expected_score) 
   %L,C,COV and expected_score are vector of size 22 by 1
   %x is the population (coefficients) that the GA uses

   score=sqrt(power(((power(L,x(1)) + power(C,x(2))+power(COV,x(3)))- 
      expected_value),2));
end

 FitnessFunction =@(x) Fitness_fct(x,L ,C,COV, expected_score );
 [x,fval] = ga(FitnessFunction,4);
 % here x will be the optimized coefficients and fval will be the   
   value obtained by the optimized coefficients

获得优化系数 (x(1), x(2),x(3)) 后，我将计算我希望它接近向量 "Expected_value" 的最终分数：

for i=1:22
final_score(i)= ((power(L,x(1)) + power(C,x(2))+power(COV,x(3));
end

我在方法中有什么地方错了吗？

Answer 1

错误消息准确说明了问题所在：您的适应度函数应该 return 是一个标量。根据您的代码，我认为您想使用某些数量的 root-mean-square 作为适应度。要实现这一点，只需 sum() 平方根即可：

function score= Fitness_fct(x,L,C,COV,expected_score) 
   %L,C,COV and expected_score are vector of size 22 by 1
   %x is the population (coefficients) that the GA uses

   score=sqrt(sum(power(((power(L,x(1)) + power(C,x(2))+power(COV,x(3)))- 
      expected_value),2))); % note the added sum()
end

 FitnessFunction =@(x) Fitness_fct(x,L ,C,COV, expected_score );
 [x,fval] = ga(FitnessFunction,4);
 % here x will be the optimized coefficients and fval will be the   
   value obtained by the optimized coefficients

Fitness 函数必须 return 一个标量

The Fitness function must return one scalar

matlab

genetic-algorithm