AI Paper Atlas

Research map / Label-efficient and robust learning / Knowledge distillation

Bi-Spectrum Distillation: Addressing Spectral Mismatch in ANN-SNN Knowledge Transfer

Yuxuan Zhang, Yuhang Sun, Wen Yao, Yue Deng, Hongjue Li

AAAI 2026 · Label-efficient and robust learning › Knowledge distillation · approach: Rethinking distillation objectives · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add SRD (feature-level spectral residual) to logits KD (ablation)SEW-ResNet-19 · CIFAR-100Top-1 accuracy (%): 79.46 -> 82.52Improved
add SSD (sorted-logit high-frequency distillation) to logits KD (ablation)SEW-ResNet-19 · CIFAR-100Top-1 accuracy (%): 79.46 -> 79.98Improved
add SSD to logits KD (ablation)SEW-ResNet-19 · CIFAR-10Top-1 accuracy (%): 96.33 -> 96.57Improved
add SRD to logits KD (ablation)SEW-ResNet-19 · CIFAR-10Top-1 accuracy (%): 96.33 -> 96.53Improved
combine SRD and SSD (full BSD) versus SRD alone (ablation)SEW-ResNet-19 · CIFAR-100Top-1 accuracy (%): 82.52 -> 82.61Improved
add standard logits KD over no distillation (ablation)SEW-ResNet-19 · CIFAR-100Top-1 accuracy (%): 77.29 -> 79.46Improved
apply SRD to all three blocks instead of a single block (ablation)ResNet-19 · CIFAR-10Top-1 accuracy (%): 95.87 -> 97.38Improved
use FFT-domain SRD instead of spatial-only adapter (ablation)SEW-ResNet-18 · ImageNetTop-1 accuracy (%): 65.47 -> 66.92Improved
apply BSD distillation on top of Sformer-12-512 initialized from E-SpikeFormerSformer-12-512 · ImageNetTop-1 accuracy (%): 83.20 -> 85.33Improved
apply BSD on Sformer-4-384 versus BKDSNNSformer-4-384 · CIFAR-100Top-1 accuracy (%): 81.26 -> 83.51Improved
BSD versus TLD on ResNet-19ResNet-19 · CIFAR-100Top-1 accuracy (%): 82.47 -> 82.36No clear effect
SNN-to-SNN BSD distillation from T=16 teacher to T=10 studentSformer-2-256 · CIFAR10-DVSTop-1 accuracy (%): 79.80 -> 81.20Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy