I’m Naomi, I’m starting the Practical Deep Learning Part 1 today and I’d like to know if could I implement a K-Means algorithm using Neural Networks, but mainly to know if I could get better results compared to the algorithms we already have.
Are you thinking rather than finding the new mean you would produce a proposed centroid mean of weights * historically proposed means? How would you know the final centroids were optimal. How are the weights learnt.
Hi, @Conwyn, thanks for your reply.
I’m actually thinking of something more general, like the algortithms we see i.e., from scikit-learn.