Pith. sign in

REVIEW 3 cited by

Decoupled Kullback-Leibler Divergence Loss

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2305.13948 v3 pith:7VW4KQZC submitted 2023-05-23 cs.CV cs.LG

Jiequan Cui , Zhuotao Tian , Zhisheng Zhong , Xiaojuan Qi , Bei Yu , Hanwang Zhang This is my paper · ORCID
classification cs.CVcs.LG
keywords lossdivergencekullback-leiblerdistillationknowledgeadversarialdecoupledtraining
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
abstract

In this paper, we delve deeper into the Kullback-Leibler (KL) Divergence loss and mathematically prove that it is equivalent to the Decoupled Kullback-Leibler (DKL) Divergence loss that consists of 1) a weighted Mean Square Error (wMSE) loss and 2) a Cross-Entropy loss incorporating soft labels. Thanks to the decomposed formulation of DKL loss, we have identified two areas for improvement. Firstly, we address the limitation of KL/DKL in scenarios like knowledge distillation by breaking its asymmetric optimization property. This modification ensures that the $\mathbf{w}$MSE component is always effective during training, providing extra constructive cues. Secondly, we introduce class-wise global information into KL/DKL to mitigate bias from individual samples. With these two enhancements, we derive the Improved Kullback-Leibler (IKL) Divergence loss and evaluate its effectiveness by conducting experiments on CIFAR-10/100 and ImageNet datasets, focusing on adversarial training, and knowledge distillation tasks. The proposed approach achieves new state-of-the-art adversarial robustness on the public leaderboard -- RobustBench and competitive performance on knowledge distillation, demonstrating the substantial practical merits. Our code is available at https://github.com/jiequancui/DKL.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Implicit Inversion turns CLIP into a Decoder

    cs.CV 2025-05 conditional novelty 6.0 of 10

    A frozen CLIP encoder can be inverted through frequency-aware implicit neural representations to generate images, reconstruct photos, edit scenes, and transfer style, all without training a decoder.

  2. Wasserstein distributional adversarial training for deep neural networks

    cs.LG 2025-02 conditional novelty 5.0 of 10

    Extends TRADES to a Wasserstein distributional threat with an efficient budgeted attack, and fine-tunes pre-trained robust models to improve a self-defined distributional robustness metric.

  3. Understanding Adversarial Training with Energy-based Models

    cs.LG 2025-05 conditional novelty 4.0 of 10

    Delta energy, the energy gap between an image and its adversarial counterpart, separates catastrophic from robust overfitting, and penalizing it with the DER regularizer mitigates both while improving generation diversity.

Pith tools