论文部分内容阅读
根据DNA微阵列数据维数高、样本少的特点,提出了一种用于癌症分子分型的演化硬件方法.该方法是基于虚拟可重构结构的演化硬件识别方法,具备灵活、高速和自适应能力强的特点,有利于建立一个拥有高数据吞吐能力、学习结果易读的高效分类系统.采用信噪比的方法进行特征选择,对选择的基因经过归一化和二值化,然后用演化硬件识别系统通过系统学习和系统分类2个阶段进行处理.急性白血病分类的硬件实验结果表明:演化硬件的识别率和识别时间分别达到了95.88%和0.12μs.
According to the characteristics of DNA microarray with high dimension and few samples, a novel evolutionary hardware method for cancer molecular typing is proposed.The method is based on the evolutionary hardware identification method of virtual reconfigurable structure, which has the advantages of flexible, high-speed and self- Adaptability of the characteristics conducive to the establishment of a high data throughput, learning results easy to read high-efficiency classification system using signal to noise ratio method for feature selection, the selected gene is normalized and binarized, and then use The evolution hardware recognition system is processed by system learning and system classification in two stages.The experimental results of hardware classification of acute leukemia show that the recognition rate and recognition time of evolution hardware reach 95.88% and 0.12μs, respectively.