基于模糊密度峰值聚类的区域同位模式并行挖掘算法

区域同位模式挖掘(RCPM,regional co-location pattern mining)是为了发掘某个局部区域内存在的同位(co-location)模式,以发现在全局中无法发现的信息.传统的区域挖掘大多会采用明确界限的几何体框定同位模式产生的区域.但是现实中的各类区域可能是无明确边界的.另外,数据的分布情况作为区域的重要特征之一,也应该成为区域选择的因素.基于上述思考,本文引入密度峰值聚类(DPC,density peak-based clustering),提出新的密度度量函数,并结合模糊集理论与k近邻距离,设计了一个行之有效的并行区域同位模式挖掘算法.实验结果表明,利用本文方法挖掘到的结果更具有现实意义,并且并行化极大地提升了挖掘算法的效率.在真实数据上,2线程下的加速比达到了1.89.

Radiation dose rate estimates to marine biota resulting from the routine radioactive releases of the nuclear power plants in Daya Bay

Five marine organisms in Daya Bay are selected for estimating the dose rates. Internal exposure and its 5 pathways are also considered. The concentrations of 43 kinds of background and additional radionuclides in seawater are from the survey of background radioactivity and the prediction of routine radioactive releases of the 4 reactors. Point source dose distribution functions are used to estimate the dose rates to various organisms from a and β radiation.