FPGA implementation of extreme learning machine system for classification
Tan Chong Yeam, Nordinah Binti Ismail, Koichiro Mashiko, Takanori Matsuzaki · 2017
The availability of intelligent embedded system to assist the classification application is a great challenge in machine learning field in last few decades. Extreme Learning Machine (ELM) is one of the best learning methods for the implementation due to its classification accuracy and speed. The main computational effort of ELM is to compute the pseudo-inverse of hidden layers output. This work presents a Modified Gram-Schmidt QR decomposition (MGS-QRD) method and hardware architecture for the FPGA implementation of ELM system. The proposed algorithm is implemented on MATLAB and compared with ordinary ELM computational methods. Next, an embedded ELM system is presented, where the software part is developed using C language and hardware is created on field-programmable gate array (FPGA). Experimental results show that the proposed algorithm achieve better performance than original ELM computational method. Finally, the investigation is performed by altering the activation function and number of hidden neurons, providing a guideline on selecting the best parameter for classification.