论文部分内容阅读
用户对个体的评价和用户满意度之间的关系是影响交互式进化计算(IEC)全局收敛性的重要因素.首先,基于用户对个体评价和用户满意度占优关系,把 IEC 中的用户分为4类:绝对理性用户、有限理性用户、有限非理性用户和绝对非理性用户.其次,给出关于 IEC 全局收敛的4个定理及理性用户是 IEC 全局收敛的充分条件这一结论,并指出 IEC 的全局收敛需要保留两个最优:适应值最优和满意度最优.最后,通过实验进一步验证上述结论.结论表明,在 IEC 中,当其它保证算法收敛的条件具备时,用户只要保证理性条件,就能保证算法全局收敛.
The relationship between user’s individual evaluation and user’s satisfaction is an important factor affecting the global convergence of Interactive Evolution Computing (IEC) .Firstly, based on the dominant relationship between individual evaluation and user satisfaction, Are four types: absolute rational user, bounded rational user, bounded irrational user and irrational irrational user.Secondly, given the four theorems about IEC global convergence and rational users’ sufficient conditions for IEC global convergence, we point out The global convergence of IEC needs to retain two optimal values: optimal fitness and optimal satisfaction.Finally, the above conclusion is verified by experiments.The conclusion shows that in IEC, when the conditions for convergence of other guaranteed algorithms are satisfied, the user only needs to ensure Rational conditions, we can ensure the global convergence of the algorithm.