A Hybrid Algorithm of Extreme Learning Machine and Sparse Auto-Encoder

This paper proposes two deep learning modes which combine sparse auto-encoder with extreme learning machine (ELM) and kernel extreme learning machine (KELM), namely as Stacked Sparse Auto-encoder-Extreme Learning Machine and Stacked Sparse Auto-encoder-Kernel Extreme Learning Machine. The proposed models are applied to the image recognition task. To learn features from the original input data, SSAE with deep architecture is employed. Then, to construct a unified neural network learning model, ELM and KELM are selected to classify the extracted features. To evaluate the performance of the two proposed models, we carry out experiments on three different image data sets respectively. The results show that the proposed models’ performance can be not only superior to the shallow architecture models of Support Vector Machine, ELM and KELM but also better than the deep architecture models, such as Stacked Auto-encoder, Deep Belief Network and Stacked Denoising Auto-encoder.