基于超限学习机的快速癌症检测方法

利用基于局部感受野的超限学习机(ELM-LRF)算法从给定的基因表达数据中提取有效的特征来进行癌症检测与分类.首先使用主成分分析（PCA）方法对原数据进行适当预处理，减少数据中存在的冗余，然后构建特定的特征映射，将得到的数据映射到相应特征空间中去，最后对得到的数据特征进行训练学习，得到最终训练好的特征提取模型. 实验表明，ELM-LRF的学习效率更高，取得的癌症检测效果比以往方法更好.
Abstract：The local receptive fields based extreme learning machine (ELM-LRF) method was utilized to learn the effective features from the acquired gene expression data to help enhance cancer diagnosis and classification. Firstly, the principal component analysis (PCA) method was implemented to process the dataset. Secondly, the features mapping to map our dataset were constructed to the specific feature space. Finally, the features to train the learning model were used to get the final ELM feature extraction model. The experiment shows that the proposed algorithm outperforms almost all the existing methods in accuracy and efficiency.