Class-Level Logit Perturbation

2022-09-13 · Mengyang Li, Fengguang Su, Ou wu, Ji Zhang

Features, logits, and labels are the three primary data when a sample passes through a deep neural network. Feature perturbation and label perturbation receive increasing attention in recent years. They have been proven to be useful in various deep learning approaches. For example, (adversarial) feature perturbation can improve the robustness or even generalization capability of learned models. However, limited studies have explicitly explored for the perturbation of logit vectors. This work discusses several existing methods related to class-level logit perturbation. A unified viewpoint between positive/negative data augmentation and loss variations incurred by logit perturbation is established. A theoretical analysis is provided to illuminate why class-level logit perturbation is useful. Accordingly, new methodologies are proposed to explicitly learn to perturb logits for both single-label and multi-label classification tasks. Extensive experiments on benchmark image classification data sets and their long-tail versions indicated the competitive performance of our learning method. As it only perturbs on logit, it can be used as a plug-in to fuse with any existing classification algorithms. All the codes are available at https://github.com/limengyang1992/lpl.

📄 PDF Abstract BibTeX arXiv:2209.05668

Code (1)

limengyang1992/lpl 공식 구현 pytorch

Tasks

Data Augmentation image-classification Image Classification Multi-Label Classification MUlTI-LABEL-ClASSIFICATION

Learning to Perturb Hidden Representations for Generalizable Deep Learning

2026-05-28 · Hua Li arxiv

Deep neural networks process data through a cascade of representations: input features, hidden activations, logits, and loss. While perturbations at the input, logit, and label levels have been systematically studied, th…

Domain Generalization

Gradient Perturbation: Learning to Perturb Gradients for Adaptive Training

2026-05-28 · Hua Li arxiv

Deep neural network training involves both forward propagation (from features through logits to loss) and backward propagation (from loss through gradients to parameter updates). While perturbations along the forward cha…

Mitigating Spurious Correlations with Causal Logit Perturbation

2025-05-21 · Xiaoling Zhou, Wei Ye, Rui Xie, Shikun Zhang

Deep learning has seen widespread success in various domains such as science, industry, and society. However, it is acknowledged that certain approaches suffer from non-robustness, relying on spurious correlations for pr…

counterfactual Long-tail Learning Meta-Learning

Understanding the Logit Distributions of Adversarially-Trained Deep Neural Networks

2021-08-26 · Landan Seguin, Anthony Ndirango, Neeli Mishra, SueYeon Chung 외

Adversarial defenses train deep neural networks to be invariant to the input perturbations from adversarial attacks. Almost all defense strategies achieve this invariance through adversarial training i.e. training on inp…

Adversarial Robustness

SWAP: Exploiting Second-Ranked Logits for Adversarial Attacks on Time Series

2023-09-06 · Chang George Dong, Liangwei Nathan Zheng, Weitong Chen, Wei Emma Zhang 외

Time series classification (TSC) has emerged as a critical task in various domains, and deep neural models have shown superior performance in TSC tasks. However, these models are vulnerable to adversarial attacks, where …

Time Series Time Series Classification