This paper analyzes 2 critical problems in FPGA-based implementation of SOM neural network algorithm: parallelizability and finite word-length effect.
分析了SOM神经网络算法在FPGA实现过程中要考虑的2个主要问题：并行性和有限字长效应。
Because word-length of register is finite. It is necessary to consider the question about the finite word-length effect during the devising digit filter.
由于寄存器的字长是有限的，因此要在数字滤波器的设计过程中考虑各种有限字长效应问题。
A new analysis method based on theoretical statistic models was proposed to analyze finite word-length effect, instead of the complicated and time-consuming sample statistics process.
理论统计模型分析的新方法的基础上，提出了有限字长效果进行分析，而不是复杂和费时的样品统计过程。
A new analysis method based on theoretical statistic models was proposed to analyze finite word-length effect, instead of the complicated and time-consuming sample statistics process.
理论统计模型分析的新方法的基础上，提出了有限字长效果进行分析，而不是复杂和费时的样品统计过程。
应用推荐