【摘要】 文中针对在文本处理的高维矢量环境中Kohonen自组织特征映射神经网络的计算瓶颈问题进行分析,引入RM(随机映射)方法并进行相应的理论分析,在此基础上提出可以运用RM方法有效并且可控地解决上述计算瓶颈问题,降低了文本处理环境中Kohonen神经网络的规模和时间、空间代价。文章通过实验证明了上述方法的有效性和正确性,从而达到提高自组织理论对于文本处理的实时性和实际可行性的目的,并对其进一步应用进行展望。更多还原

【Abstract】 This paper analyzes the bottleneck problems of calculation in Kohonen self-organizing map neutral network(SOM) in the high-dimensional vector environment of text processing,and introduces RM(Random Mapping) to make corresponding theoretic analysis,on the basis of which it suggests that RM may be employed to settle above bottleneck problems of calculation in an efficient and controllable way and to reduce size and cost of time and space for Kohonen neural network in text processing environment. This paper demonstrates the efficiency and correctness of the present method by using an example,which makes SOM network to possess real-time ability and feasibility in applying it to the text processing,and offers a prospect of its further applications.更多还原