In this paper, we address M-QAM blind equalization by fitting the probability density functions (pdf) of the equalizer output with the constellation symbols. We propose two new cost functions, based on kernel pdf approximation, which force the pdf at the equalizer output to match the known constellation pdf. The kernel bandwidth of a Parzen estimator is updated during iterations to improve the convergence speed and to decrease the residual error of the algorithms. Unlike related existing techniques, the new algorithms measure the distance error between observed and assumed pdfs for the real and imaginary parts of the equalizer output separately. The proposed approaches outperform CMA and classical pdf fitting methods in terms of convergence speed and residual error.