q-Sine Circular Extreme Learning Machine for High Dimensional Data

Circular Extreme Learning Machine(CELM) is the fast training neuron network that can handle the high dimensional data. The study of CELM has shown the effectiveness of using sine activation to improve the CELM more accuracy than the original CELM with the sigmoid function. However, the sine function can develop to the q-sine function that can make flexible to the sine activation by taking the parameters. CELM with the q-sine activation called q-Sine Circular Extreme Learning Machine (QSCELM). QSCELM is CELM with q-sine activation that uses the parameter p and q to change the amplitude and wavelength of the sine wave to get better accuracy. The experimental results show that QSCELM is more accurate than several original and applied ELM and CELM.