欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
针对目标的决策问题,依据动态模糊逻辑理论,提出一种视觉不变特征的融合算法。在多特征空间训练多个神经网络,根据各个特征空间对最终结果的贡献程度,对投射到每个特征空间所得到的特征向量计算隶属度,并进行动态模糊加权融合处理,以实现机器对特征选择的自适应性,减少人工干预并最终实现机器智能的目标。实验结果表明,该方法在特征信息缺失或含有噪声的情况下,较传统的神经网络方法有明显的提高,能达到较为满意的结果。
英文摘要：
In order to deal with the target decision problem,a fusion algorithm characterized by visual constancy is proposed based on the theory of dynamic fuzzy logic.The first thing to do is to train multiple neural network in the multi-feature space.According to the degree of each feature subset＇s contribution to the final result,the degree of membership of each feature vector obtained from each feature space is computed,and each feature vector is processed with dynamic fuzzy weight so as to enhance the adaptability of the machine to conduct feature selection and reduce the manual intervention to realize the intelligent goal of the machine.As shown in the experiment,compared with the traditional neural network algorithm,the proposed algorithm performs well under the conditions of missing feature information and noise.
同期刊论文项目
同项目期刊论文
期刊信息
位置：