For feature selection of multiclass data, Ramaswamy et al. used an one-versus-all strategy to convert the multiclass problem into a series of binary class problems and applied SVM-RFE to each binary class problem separately (Ramaswamy et al., 2001. Fu and Fu-Liu (2005) also proposed a method based on the one-versus-all strategy.
For the first one, Liu et al. (2012) proposed a method based on fuzzy K-harmonic means which could reduce the sensitivity to the initial centers to a certain extent; however, the local optimal value cannot be solved completely as the algorithm still uses the iterative optimization technique which is similar to the K-mean method.
To achieve maximized efficiency, this work proposed a method based on error compensation and minimization.
We have recently proposed a method based on finite-difference schemes to solve partial differential equations in an asynchronous fashion – synchronization between PEs is relaxed at a mathematical level.
Goljan and Fridrich [10] proposed a method based on DCT.
Brug proposed a method based on calculation of the correlation between two images [9].
Osareh [8] proposed a method based on template matching for localizing the center of optic disc.
Artigas et al.[38] have proposed a method based on depth maps.
Zio, et al [106], proposed a method based on fuzzy similarity analysis to estimate system RUL.
Fu et al. [6] proposed a method based on two stages.
Focusing on the speed, Lowe [10] proposed a method based on Difference of Gaussian (DoG) for scale invariant feature detection.
