A new study introduced mixed kernel functions for support vector machines in pattern recognition. By combining Gaussian and Wavelet functions, as well as Gaussian and Polynomial kernel functions, the researchers found that these new mixed kernels improved time efficiency and accuracy in image recognition tasks. Specifically, the mixed kernel function of Gaussian and Wavelet was identified as particularly effective for recognizing different states in images.