Thank you very much for your inputs.
I will follow those steps.
I have the following two questions.
1. when I don't have an edge between two nodes, in my adjacency matrix I will represent it as 0 or +inf?(+ inf- because as you said I have to treat it as a distance mattix right)
2. This is a follow-up question.
Now what if I have a feature vector of D dimensions at each node. In that case can you suggest me some ways of thinking about filtration. This is a common occurrence in modern deep learning applications on graphs such as graph convolutional networks.