Author : Amit R. Chavan 1
Date of Publication :21st February 2018
Abstract: This paper elaborates the implementation of an unsupervised Artificial Neural Network (ANN) on FPGA hardware for data classification. ANN is the best option to classify a large amount of data into several desired classes as per the characteristics and parameters of the given data samples. Implementation of an unsupervised ANN on a chip eliminates the additional stage of software simulation of the ANN for the given dataset, i.e. training of ANN using a software and then implementation of trained ANN on FPGA chip. The Unsupervised ANN is implemented on Xilinx Virtex-4 FPGA, which consumes less on-chip resources, consuming less power at optimum speed
Reference :
-
- Omondi, Rajapakshe, FPGA Implementations of Neural Networks, Springer.
- Brownlee, Master Machine Learning Algorithms.
- Ayman Youssef, Mohammed, Nasar, “Two Novel Generic, Reconfigurable Neural Network FPGA Architectures”, IEEE 4th International Conference on Artificial Intelligence with Applications in Engineering and Technology, 2014.
- Dinu, Cirstea, “A Digital Neural Network FPGA Direct Hardware Implementation Algorithm”, IEEE International Symposium on Industrial Electronics (ISIE), 2007.
- Kanwal, Generalized Functions: Theory and Technique, 2nd ed. Boston, MA: Birkhäuser, 1998.
- Chaitra, “Hardware Implementation of Artificial Neural Networks using Back Propagation Algorithm on FPGA”, International Journal of Research in Engineering and Technology, Vol. 05 Sp. Issue 04, pp. 211-214, May 2016.