PROJECT TITLE :
Multi-View Missing Data Completion - 2018
A growing variety of multi-view data arises naturally in many eventualities, including medical diagnosis, webpage classification, and multimedia analysis. A challenge in learning from multi-view knowledge is that not all instances are fully represented in all views, resulting in missing read data. In this Project, we specialise in feature-level completion for missing view of multi-read data. Aiming at capturing each semantic complementarity and identical distribution among totally different views, an Isomorphic Linear Correlation Analysis (ILCA) technique is proposed to linearly map multi-read data to a feature-isomorphic subspace through learning a set of fantastic isomorphic options, thereby unfolding the shared data from completely different views. Meanwhile, we have a tendency to assume that missing view obeys traditional distribution. Then, the missing view data matrix can be modeled as an occasional-rank element plus a sparse contribution. Thus, to accomplish missing view completion, an Identical Distribution Pursuit Completion (IDPC) model based mostly on the learned features is proposed, in that the identical distribution constraint of missing read to the other obtainable one within the feature-isomorphic subspace is fully exploited. Comprehensive experiments on many multi-read datasets demonstrate that our proposed framework yields promising results.
Did you like this research project?
To get this research project Guidelines, Training and Code... Click Here