Furthermore, we propose a novel matrix completion method based on multi-Gaussian models to estimate the missing traffic data.
Our approach is a novel application of the inductive matrix completion method; it can be applied to diseases not seen at training time, unlike traditional matrix completion approaches and other network-based inference methods that are transductive.
By matrix completion method [ 28– 30], we replace solving low-rank problem with dealing with nuclear norm [ 31]; then problem (7) can be rerepresented as (8) min Z || Z || ∗, s.
In this article, we apply a novel matrix-completion method called Inductive Matrix Completion to the problem of predicting gene-disease associations; it combines multiple types of evidence (features) for diseases and genes to learn latent factors that explain the observed gene disease associations.
For all the matrix-completion-based methods (including LEML and IMC), we rank the predictions using the estimated values of the matrix (higher the estimated value P ij, more relevant is the gene i for disease j).
The three competitive methods Katz, C atapult and matrix completion on the combined network which use the same information, albeit in different ways, perform very similarly within the top-100 predictions.
SDPA-C is a primal dual interior-point method using the positive definite matrix completion technique by Fukuda et al., and it performs effectively with SDPs with a large scale matrix variable, but not with a large number of equality constraints.
For example, a predictive control strategy for wide-area damping control was presented in [28] with the consideration of data loss and other physical constraints, and a data reconstruction method using the low-rank matrix completion approach was provided in [29], in which way the lost data could be partially recovered at a control center.
Nowadays methods such as collaborative filtering or matrix completion are developed to predict the missing part.
However, it is in general difficult to evaluate the prediction accuracy for methods such as collaborative filtering or matrix completion.
