What is the computational complexity of SVM training?

SVM training is typically O(n²) to O(n³) depending on solver and kernel.

Kernel SVM requires computing and storing pairwise kernel matrix, leading to quadratic memory and cubic time complexity in worst-case optimization. Linear SVM reduces complexity significantly.