Multi-View Classification via Reweighted Regression and Joint Sparsity Regularization

Multi-view data includes information from different perspectives or data sources, which is comprehensive but often contains redundant information in practical applications. This redundancy makes it challenging to efficiently use multi-view data for classification. To address this, we have developed a new multi-view classification model based on reweighted regression and joint sparsity regularization. Therefore, we propose a novel multi-view classification model based on reweighted regression and joint sparsity regularization. This model uses a non-squared F-norm in the loss function to mitigate the impact of outliers and incorporates an <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{L}_{2,1}$</tex>-norm sparsity regularization term to reduce redundant and noisy information. Unlike traditional methods, our model calculates loss terms separately in different view subspaces and balances the importance of different views. Comparative experimental results with related classification methods demonstrate the superiority of this model.