I am trying to minimize (optimize) the distance between a generated subject-specific hypergraph and another set of hypergraphs (represents another set of random subjects), what kind of loss function can I use on incidence matrices knowing that each node has a corresponding feature vector. (and each hypergraph represents a subject)
Saeid Alirezazadeh
You can use the Frobenius norm to find differences between graph structures and use the usual Euclidean distance to find the difference between feature vectors of similar nodes (if you consider any at the moment). Now you can use a linear combination of these two based on the importance of these two metrics to generate a new loss function.
0 votes
0 thanks
- Badges
- Science topic
- Similar topics
- Methodology
- Research Methodology
- Research Papers
More Mayssa Soussia's questions See All
Similar questions and discussions