TY  - EJOU
AU  - Zhang, Dongmin 
AU  - Sun, Chao 
AU  - Zhang, Yikun 
AU  - Yin, Runyao 
AU  - Chen,  

TI  - A Multi-Tree Genetic Programming Framework for Multi-View Feature Construction and Fusion
T2  - Computers, Materials \& Continua

PY  - 
VL  - 
IS  - 
SN  - 1546-2226

AB  - Multi-view tabular data poses challenges due to heterogeneous feature spaces, divergent distributions, and implicit cross-view relationships. Traditional methods further struggle because their hand-crafted fusion strategies cannot adequately capture complex nonlinear interactions among views. To address this issue, this paper proposes a genetic programming (GP)-based multi-view feature fusion method. It employs a multi-tree GP framework: each view is assigned a dedicated tree for intra-view feature selection and construction, and a fusion tree combines their outputs for cross-view feature-level fusion. An enhanced feature construction strategy further enriches the final representation by exploiting subtree information. Together with decision-level fusion, GP-based multi-view feature fusion framework (GPMVF) forms a two-level fusion framework that jointly handles intra-view construction, cross-view fusion, and classification. Experiments on eight public multi-view datasets show that the proposed method achieves better performance than the compared baseline method.
KW  - Genetic programming; multi-view learning; feature construction; feature fusion

DO  - 10.32604/cmc.2026.086570