Learning Auto-Structured Regressor from Uncertain Nonnegative Labels

In this paper, we take the human age and pose estimation problems as examples to study automatic designing regressor from training samples with uncertain nonnegative labels. First, the nonnegative label is predicted as the square norm of a matrix, which is bilinearly transformed from the nonlinear mappings of the candidate kernels. Two transformation matrices are then learned for deriving such a matrix by solving a semidefinite programming (SDP) problem, in which the uncertain label of each sample is expressed as two inequality constraints. The objective function of SDP controls the ranks of these two matrices, and consequently automatically determines the structure of the regressor. The whole framework for automatic designing regressor from samples with uncertain nonnegative labels has the following characteristics: 1) SDP formulation makes full use of the uncertain labels, instead of using conventional fixed labels; 2) regression with matrix norm naturally guarantees the nonnegativi...

• Type: Conference
• Year: 2007
• Where: ICCV
• Authors: Shuicheng Yan, Huan Wang, Xiaoou Tang, Thomas S. Huang
• Keywords: Automatic Designing Regressor, Computer Vision, Conventional Fixed Labels, ICCV 2007, Pose Estimation Problems, Uncertain Labels, Uncertain Nonnegative Labels