Can you explain the concept of kernel techniques in machine learning?
What conditions must a matrix meet to qualify as a kernel?
What are the consequences of using a kernel that does not meet these conditions when running a support vector machine model?