is backpropagation used in CNN for learning filters?
Cnn learning filters
Yes Salik, Backpropagation is indeed used to learn the weights of Convolutional Kernels/Filters. As a matter of fact, all the weights in any kind of layer in modern Deep Learning (except for some exceptions like Capsule Layers, Swarm Optimization, L-BFGS) are learnt through Backprop and Gradient Descent.
If you are interested in the Maths behind Backprop in CNNs, checkout the following links: