【摘要】 为了解决难以利用 MDS方法进行有效的增量式数据索引的问题 ,本文提出了基于神经网络的增量式数据索引方法 ,该方法首先对少量数据用 MDS方法进行索引 ,索引结果用于训练神经网络 ,新数据再通过训练后的神经网络进行索引 .利用训练后的神经网络对增量数据进行索引的时间复杂度为 O(n) .实验结果表明 ,本文防哪个法可以有效的进行增量式数据索引 ,并交好的保持了数据对象间的距离信息更多还原

【Abstract】 To solve the problem that MDS cannot effectively be used for incremental data index, the paper proposes a neural network based incremental data indexing approach. In this approach, first a small number of data is indexed with MD, and the index is used to train neural network. New data then can be indexed with the trained neural network. The time complexity of indexing data with trained neural network is O(n). Our experimental results show our method can effectively perform incremental data indexing, and can keep well the distance information between data objects.更多还原