Review on Methods to Fix Number of Hidden Neurons in Neural Networks

Table 5

Performance analysis of various approaches in existing and proposed models.

S. no.	Various methods	Year	Number of hidden neurons	MSE
1	Li et al. method [7]	1995	𝑁ℎ=√1+8𝑛−1/2	0.0399
2	Tamura and Tateishi method [9]	1997	𝑁ℎ=𝑁−1	0.217
3	Fujita method [10]	1998	𝑁ℎ=𝐾log𝑃𝑐𝑍/log𝑆	0.0723
4	Zhang et al. method [14]	2003	𝑁ℎ=2^𝑛/𝑛+1	0.217
5	Jinchuan and Xinzhe method [3]	2008	𝑁ℎ=𝑁in+𝑁𝑝/𝐿	0.0299
6	Xu and Chen method [19]	2008	𝑁ℎ=𝐶𝑓𝑁/𝑑log𝑁^0.5	0.0727
7	Shibata and Ikeda method [20]	2009	𝑁ℎ=𝑁𝑖𝑁0	0.1812
8	Hunter et al. method [2]	2012	𝑁ℎ=2^𝑛−1	0.0727
9	Proposed approach		𝑁ℎ=4𝑛^2+3/𝑛^2−8	0.018