SeAl-KD selectively aligns class-level and temporal knowledge in SNNs by equalizing competing logits at erroneous timesteps and reweighting alignment by confidence and inter-timestep similarity, outperforming uniform KD methods on image and event datasets.
Head-tail-aware kl divergence in knowledge distillation for spiking neural networks.arXiv preprint arXiv:2504.20445
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Not All Timesteps Matter Equally: Selective Alignment Knowledge Distillation for Spiking Neural Networks
SeAl-KD selectively aligns class-level and temporal knowledge in SNNs by equalizing competing logits at erroneous timesteps and reweighting alignment by confidence and inter-timestep similarity, outperforming uniform KD methods on image and event datasets.