cs.CVNov 7, 2026

Heterogeneous Asymmetric Contrastive Distillation

Authors: Qi Wang, Jinjia Zhou

Organizations: Graduate School of Science and Engineering, Hosei University, 3-7-2 Kajino-cho, Koganei-shi, Tokyo, 184-8584, Japan

Abstract

Knowledge distillation (KD) between architectural families, e.g., from a Transformer teacher to a CNN student, is difficult because the two networks learn different representations. Logits share a label space, but heterogeneous teachers differ in logit offset, scale, and sharpness, and their non-target classes are less semantically organized. Conventional logit distillation matches these values sample by sample, asking the student to imitate how another architecture expresses its predictions. We argue that a heterogeneous teacher should pass on the relational structure of its predictions, i.e., how it arranges samples relative to one another, and propose Heterogeneous Asymmetric Contrastive Distillation (HACD) to transfer this structure. HACD maps teacher and student logits to centered, temperature-softened probability directions on a shared unit hypersphere, which removes architecture-dependent offsets and lets the temperature control how sharply peaked outputs enter the representation while keeping informative non-target classes. Each student output must then identify its own teacher target within the batch. Because every teacher target also acts as a negative for the others, HACD computes the targets on a moderately augmented view and contrasts both a moderate and a strong student view against them. HACD adds no learnable parameters and needs one teacher pass per sample. It achieves the best or second-best accuracy on 11 of 12 heterogeneous pairs on CIFAR-100, without ground-truth labels, and on 14 of 15 pairs on ImageNet-1K, improving over the reported vanilla KD results by 8.82 and 1.33 percentage points on average. Checkpoints are available at https://github.com/BTAEYANG/HACD-Heterogeneous-Asymmetric-Contrastive-Distillation; the code will be released upon publication.

Figures & tables

Explore similar work

Oct 11, 2026

Representational Difference Distillation

Knowledge distillation transfers representational knowledge from a large teacher network to a compact student. Existing relational and contrastive distillation methods treat all samples uniformly, spending equal gradient on well-matched and poorly-matched regions of the representation space alike. We propose Representational Difference Distillation (RDD), which diagnoses where teacher and student representations structurally disagree and concentrates the distillation signal on those regions. RDD uses two asymmetric affinity matrices that capture opposing directions of misalignment: one identifies samples the student incorrectly considers similar, the other identifies teacher relationships the student has yet to learn. These affinities determine which sample pairs the distillation loss aligns and how strongly it corrects each misaligned sample. On CIFAR-100 with resnet32x4 to resnet8x4, RDD reaches 75.58% test accuracy in our runs, on par with CRD (75.45%) and above KD and RKD. We additionally explore curriculum-based and sampling-based variants, finding that weighting the loss is more effective than filtering the data.
Aug 12, 2026

Anti-Shortcut Distillation via Temporal Negative Knowledge Transfer

Knowledge distillation (KD) trains a compact student by attracting it towards a converged teacher. It is silent about which directions the teacher itself learned to suppress: repulsive and bias-aware objectives exist, but none exploits the teacher's own trajectory to identify what the student should avoid. We observe that the missing signal is already encoded in the teacher's optimization trajectory: features that an early-stage teacher emphasizes but that a converged teacher attenuates are precisely the shortcut directions worth pushing the student away from. We instantiate this observation as Anti-Shortcut Distillation (ASD), a push--pull KD framework that treats the converged teacher \Tfinal\Tfinal as a positive semantic anchor and an early-checkpoint teacher \Tearly\Tearly as a temporal negative reference. ASD couples two losses: a temporal contrastive loss (\Ltc\Ltc) that places the early-teacher feature as a same-sample negative against in-batch and memory-bank final-teacher features in an InfoNCE objective; and a shortcut suppression loss (\Lss\Lss) that penalizes student projection onto the top eigenvectors of \E[\Dh\Dh⊤]\E[\Dh\Dh^{\top}], the uncentered second-moment matrix of early-to-final feature displacements. Across 13 teacher--student pairs on CIFAR-100, ImageNet-100, and TinyImageNet, ASD attains the highest clean top-1 accuracy on more than 10 pairs and outperforms standard KD on 12. On CIFAR-100-C corruption robustness, ASD obtains the lowest mean Corruption Error (86.186.1,mCE) on the most challenging cross-architecture pair (WRN-40-2→\toShuffleNet-V2). Mechanistic diagnostics confirm the intended geometry: the ASD student is systematically anti-aligned with the shortcut direction, while its projection onto the robust subspace is substantially larger (0.450.45 vs.\ 0.120.12).
Jul 29, 2026

CoCaRS: Correlation Calibration-Based Redundancy Suppression for Heterogeneous Knowledge Distillation

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.