Skip to main content
Figure 1 | BMC Systems Biology

Figure 1

From: Protein functional properties prediction in sparsely-label PPI networks through regularized non-negative matrix factorization

Figure 1

Label deficiency problem in constructing relational features x R and latent linkage solution, one approach to generate the relational features is to count the number of neighboring nodes labeled "+" or "-" (for binary problem). (a) The unknown protein is linked to 3 labeled protein where 2 are positive and 1 is negative, as such x R =< 2; 1 >. (b) The labels of neighboring nodes are initially unknown, which makes the compute of relational features very challenging; (c) By adding latent linkages, the unknown protein is connect to the most relevant labeled nodes.

Back to article page