Details

Autor(en) / Beteiligte

• Zhang, Qi
• Lu, Mingfei
• Yu, Shujian
• Xin, Jingmin
• Chen, Badong

Titel

Discovering common information in multi-view data

Ist Teil von

• Information fusion, 2024-08, Vol.108, p.102400, Article 102400

Ort / Verlag

Elsevier B.V

Erscheinungsjahr

2024

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• We introduce an innovative and mathematically rigorous definition for computing common information from multi-view data, drawing inspiration from Gács-Körner common information in information theory. Leveraging this definition, we develop a novel supervised multi-view learning framework to capture both common and unique information. By explicitly minimizing a total correlation term, the extracted common information and the unique information from each view are forced to be independent of each other, which, in turn, theoretically guarantees the effectiveness of our framework. To estimate information-theoretic quantities, our framework employs matrix-based Rényi’s α-order entropy functional, which forgoes the need for variational approximation and distributional estimation in high-dimensional space. Theoretical proof is provided that our framework can faithfully discover both common and unique information from multi-view data. Experiments on synthetic and seven benchmark real-world datasets demonstrate the superior performance of our proposed framework over state-of-the-art approaches. •We formulate a mathematically definition of common information for multiview data.•We propose a multiview learning method that discerns common and unique information.•Our method is scalable to data more than two view, supported by theoretical proof.•Experimental results substantiate the efficacy of the proposed framework.