What is too much though? Backpropagation uses derivatives, some filters in Computer Vision use multivariate calculus. If you want to have a thorough understandi

What is too much though? Backpropagation uses derivatives, some filters in Computer Vision use multivariate calculus. If you want to have a thorough understanding then calculus is necessary. That said, Andrew Ng was quite good at avoiding calculus in his Machine Learning MOOC, and for applied machine learning I guess calculus is not that important. A great place to study about math is www.khanacademy.org, they have courses on calculus, probability/statistics and linear algebra.