Missing values estimation based on a multi-view learning algorithm

Authors: Mengjiao Qin*, Zhejiang university, Zhenhong Du, Zhejiang university
Topics: Spatial Analysis & Modeling, Geographic Information Science and Systems
Keywords: missing values estimation, matrix completion, multi-view learning, ANN
Session Type: Paper
Day: 4/6/2019
Start / End Time: 9:55 AM / 11:35 AM
Room: 8217, Park Tower Suites, Marriott, Lobby Level
Presentation File: No File Uploaded


Buoy systems have been deployed to monitor coastal waters, generating massive amounts of data with high temporal resolution. However, the records are usually lost in practice, resulting in an adverse impact on further data analysis. Current imputation methods put more emphasis on the spatial and temporal correlations while ignoring the crucial interactions between different parameters. To fully use these interactions and further improve the imputation performance, a matrix completion based multi-view learning method is proposed in this paper to fill the missing values in buoy monitoring data. This method considers three hybrid views, i.e., the temporal-parameter view, the spatial-parameter view and the spatio-temporal view, and each view is formulated in a matrix. The fixed-point continuation with approximate singular value decomposition (FPCA) algorithm is utilized to reconstruct the three matrices. Then, the artificial neural network (ANN) based multi-view learning algorithm is used to aggregate the estimates of the three views into the final results. Finally, the buoy monitoring dataset of the Zhejiang coastal area is used to verify the imputation ability of the proposed model. The results confirm that the proposed model achieves better performance than other approaches.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login