LSQR: An Algorithm for Sparse Linear Equations and Sparse Least Squares
ACM Transactions on Mathematical Software (TOMS)
Laplacian Eigenmaps for dimensionality reduction and data representation
Neural Computation
Modeling interactome: scale-free or geometric?
Bioinformatics
BIBE '05 Proceedings of the Fifth IEEE Symposium on Bioinformatics and Bioengineering
ICML '06 Proceedings of the 23rd international conference on Machine learning
A lock-and-key model for protein--protein interactions
Bioinformatics
IEEE Transactions on Pattern Analysis and Machine Intelligence
Artificial Intelligence in Medicine
Random classification noise defeats all convex potential boosters
Machine Learning
Unsupervised learning of image manifolds by semidefinite programming
CVPR'04 Proceedings of the 2004 IEEE computer society conference on Computer vision and pattern recognition
Nonconvex Online Support Vector Machines
IEEE Transactions on Pattern Analysis and Machine Intelligence
The Journal of Machine Learning Research
Inferring Networks of Diffusion and Influence
ACM Transactions on Knowledge Discovery from Data (TKDD)
Hi-index | 0.01 |
Over the last decade, the development of high-throughput techniques has resulted in a rapid accumulation of protein-protein interaction (PPI) data. However, the high-throughput experimental interaction data is prone to exhibit high level of noise. Despite the promising performance of current geometric approaches for increasing the reliability of PPI networks, it is still of major concern to find a better method that requires less structural assumptions and is more robust against the large fraction of noisy PPIs. In this paper, we propose a new approach called non-convex semantic embedding (NCSE) for assessing the reliability of interactions. Unlike previous approaches which seek to preserve a predefined distance matrix in the embedding space, NCSE tries to adaptively learn a Euclidean embedding under the simple geometric assumption of PPI networks. We also propose using non-convex cost function in order to improve the robustness of NCSE. The experimental results show that our approach substantially outperforms previous methods on PPI assessment problems. NCSE could thus facilitate further graph-based studies of PPIs and may help infer their hidden underlying biological knowledge. The Matlab source code of NCSE is freely available upon request.