Search papers, labs, and topics across Lattice.
This paper introduces CoCaRS, a novel approach to knowledge distillation that enhances the transfer of knowledge between heterogeneous models by calibrating feature decorrelation to retain structural information while suppressing redundancy. By employing Confusion Evidence Estimation (CEE) and Strength Allocation Control (SAC), CoCaRS effectively captures semantic relationships and maintains discriminative structures during the distillation process. Extensive experiments demonstrate that CoCaRS significantly improves distillation performance on CIFAR-100 and ImageNet-1K while reducing sensitivity to coefficient settings across varying teacher-student pairs and training stages.
CoCaRS achieves superior knowledge transfer in heterogeneous model distillation by balancing redundancy suppression with structural integrity, outperforming traditional methods.
Knowledge distillation (KD) enables a compact student model to learn from a powerful teacher and has become an effective paradigm for model compression. The emergence of diverse model architectures has extended KD from homogeneous to heterogeneous settings. However, differences in architectural inductive biases between the teacher and student models often result in substantial representation discrepancies, limiting the effectiveness of direct knowledge transfer. Recently, redundancy suppression has offered a new perspective on heterogeneous KD by preserving cross-architecture invariance and reducing feature redundancy through decorrelation of teacher-student feature correlations. Nevertheless, this formulation may weaken useful structural information through uniform decorrelation, while a fixed coefficient may make the effective contribution of redundancy suppression sensitive to teacher-student pairs and training stages. To address these problems, Correlation Calibration-based Redundancy Suppression (CoCaRS) is proposed to better retain structural information while suppressing redundancy and reduce sensitivity to coefficient settings across teacher-student pairs and training stages. Specifically, CoCaRS calibrates feature decorrelation through Confusion Evidence Estimation (CEE) and Strength Allocation Control (SAC), which respectively capture reliable semantic relations for correlation estimation and preserve discriminative structure during decorrelation. Adaptive Coefficient Regulation (ACR) further regulates the contribution of the calibrated redundancy suppression objective according to its relative loss scale, reducing sensitivity to coefficient settings. Extensive experiments on CIFAR-100 and ImageNet-1K validate the effectiveness of CoCaRS in improving distillation performance and reducing sensitivity to coefficient settings. Code will be released soon.
