论文部分内容阅读
为了解决交互式遗传算法的用户疲劳问题,提出区间适应值交互式遗传算法神经网络代理模型.首先,对用户已评价个体的基因型及其适应值进行采样以训练神经网络,使其逼近区间适应值的上下限;然后,利用神经网络代理模型,评价后续的部分进化个体,并不断更新训练数据和代理模型,以保证逼近精度;最后,对算法性能进行了定量分析,并将其应用于服装进化设计系统.分析结果表明,所提算法在减轻用户疲劳的前提下,具有更多找到满意解的机会.
In order to solve the user fatigue problem of interactive genetic algorithm, an interactive genetic algorithm based on interval adaptive neural network neural network proxy model is proposed.Firstly, the genotypes and fitness of user-evaluated individuals are sampled to train the neural network to adapt the approximation interval Then use neural network proxy model to evaluate the subsequent evolutionary individuals, and constantly update the training data and the proxy model to ensure the approximation accuracy; Finally, the performance of the algorithm was quantitatively analyzed and applied to clothing Evolutionary design system.Analysis results show that the proposed algorithm has more opportunities to find satisfactory solutions on the premise of reducing user fatigue.