ELMDF: A new classification algorithm based on Data Field
Shuliang Wang, Dakui Wang · 2014
In this paper, a novel classification algorithm, ELMDF (Extreme Learning Machine based on Data Field), is proposed to solve the problem of estimating the number of hidden layer neurons in typical ELM. For constructing ELMDF, a new theory based on data field, FMDF (Fundamental Matrix of Data Field) is proposed in this paper. The breast cancer cell image dataset, and the genome dataset are used to test and illustrate the proposed method. The experimental case demonstrates that ELMDF performs better than other six typical supervised learning algorithms on different datasets.