generate_points: Compute the points to be separated in Step 2 of COSNet...

In COSNet: Cost Sensitive Network for node label prediction on graphs with highly unbalanced labelings

Description

This function associates each labeled node with a point in the plane, whose coordinates are respectively the weighted sum of its positive and negative neighborhoods

Usage

generate_points(W, unlabeled, labeling)

Arguments

W	square symmetric named matrix, whose components are in the [0,1] interval. The i,j-th component is the weight between node i and node j. The components of the diagonal of W are zero.
unlabeled	vector of the indices of the unlabeled nodes
labeling	vector of node labels : 1 for positive nodes, -1 for negative nodes, 0 for unlabeled nodes

Details

For each labeld node k, a point (pos_vect[k], neg_vect[k]) is computed, where pos_vect[k] is the whighted sum of the positive neighbors of node k and neg_vect[k] is the weighted sum of negative neighbors of node k.

Value

List of two element:

pos_vect	is the vector of the abscissae; pos_vect[k] contains the whighted sum of the positive neighbors of node k
neg_vect	is the vector of the ordinates; neg_vect[k] contains the whighted sum of the negative neighbors of node k

References

Frasca M., Bertoni A., Re M., Valentini G.: A neural network algorithm for semi-supervised node label learning from unbalanced data. Neural Networks, Volume 43, July, 2013 Pages 84-98.

Examples

## randomly generating labels
labels <- generate_labels(100, 0.3);
unlabeled <- sample(1:100, 10);
labels[unlabeled] <- 0;
## randomly generating connection matrix
W <- matrix(sample(1:10000, 100*100)/1000, nrow = 100);
diag(W) <- 0;
points <- generate_points(W, unlabeled, labels);
points$pos_vect[1:5];
points$neg_vect[1:5];

COSNet documentation built on Nov. 8, 2020, 8:12 p.m.