Could you discuss the computational complexity involved in training an SVM? What is the time complexity associated with making predictions using a trained SVM? How does the choice of kernel in an SVM affect the time complexity for both training and inference?

Interview question asked to Machine Learning Engineers interviewing at Wattpad, Microsoft, Bumble and others: Could you discuss the computational complexity involved in training an SVM? What is the time complexity associated with making predictions using a trained SVM? How does the choice of kernel in an SVM affect the time complexity for both training and inference?.