Virtual Adversarial Training

Virtual adversarial training (VAT) is a technique used to enhance the robustness and generalization of machine learning models, particularly deep neural networks, by training them to be invariant to small, adversarial perturbations of the input data. Current research focuses on applying VAT across diverse domains, including image classification (e.g., medical imaging, satellite imagery), natural language processing (e.g., sentiment analysis, multi-lingual emotion recognition), and 3D point cloud processing, often integrated with other techniques like contrastive learning or active learning. The improved robustness and generalization achieved through VAT contribute to more reliable and accurate models in various applications, leading to advancements in fields ranging from medical diagnosis to environmental monitoring.

Papers

September 3, 2023

A Visual Interpretation-Based Self-Improved Classification System Using Virtual Adversarial Training
Shuai Jiang, Sayaka Kamei, Chen Li, Shengzhe Hou, Yasuhiko Morimoto
Natural Language Processing Self Supervised Learning BERT Model Fine Tuned BERT Sentiment Feature Virtual Adversarial Training

December 28, 2022

Adversarial Virtual Exemplar Learning for Label-Frugal Satellite Image Change Detection
Hichem Sahbi, Sebastien Deschamps
Adversarial Attack Active Learning Adversarial Learning Change Detection Exemplar Image Virtual Adversarial Training

September 16, 2022

Hybrid Window Attention Based Transformer Architecture for Brain Tumor Segmentation
Himashi Peiris, Munawar Hayat, Zhaolin Chen, Gary Egan, Mehrtash Harandi
Brain Tumor Segmentation Multi Modal MRI MRI Volume Virtual Adversarial Training Volumetric Transformer Hybrid Attention Transformer

September 14, 2022

PointACL:Adversarial Contrastive Learning for Robust Point Clouds Representation under Adversarial Attack
Junxuan Huang, Yatong An, Lu cheng, Bai Chen, Junsong Yuan, Chunming Qiao
Adversarial Attack Adversarial Contrastive Learning 3D Adversarial Virtual Adversarial Training

May 29, 2022

Context-based Virtual Adversarial Training for Text Classification with Noisy Labels
Do-Myoung Lee, Yeachan Kim, Chang-gyun Seo
Deep Neural Network Adversarial Training Text Classification Noisy Label Virtual Adversarial Training

January 30, 2022

Co-Regularized Adversarial Learning for Multi-Domain Text Classification
Yuan Wu, Diana Inkpen, Ahmed El-Roby
Adversarial Learning Domain Alignment Domain Text Classification Domain Divergence Virtual Adversarial Training

January 25, 2022

Virtual Adversarial Training for Semi-supervised Breast Mass Classification
Xuxin Chen, Ximin Wang, Ke Zhang, Kar-Ming Fung, Theresa C. Thai, Kathleen Moore, Robert S. Mannel, Hong Liu, Bin Zheng, Yuchen Qiu
Adversarial Loss Mammogram Classification Virtual Adversarial Training

January 11, 2022

Reciprocal Adversarial Learning for Brain Tumor Segmentation: A Solution to BraTS Challenge 2021 Segmentation Task
Himashi Peiris, Zhaolin Chen, Gary Egan, Mehrtash Harandi
Adversarial Learning Segmentation Task Solution Path Brain Tumor Segmentation Brain Tumor Robust Segmentation BRAT Challenge Virtual Adversarial Training Reciprocal Learning

November 11, 2021

Multilingual and Multilabel Emotion Recognition using Virtual Adversarial Training
Vikram Gupta
Multi Label Text Classification Multi Label Emotion Cross Lingual Emotion Virtual Adversarial Training

Virtual Adversarial Training

Papers

A Visual Interpretation-Based Self-Improved Classification System Using Virtual Adversarial Training

Adversarial Virtual Exemplar Learning for Label-Frugal Satellite Image Change Detection

Hybrid Window Attention Based Transformer Architecture for Brain Tumor Segmentation

PointACL:Adversarial Contrastive Learning for Robust Point Clouds Representation under Adversarial Attack

Context-based Virtual Adversarial Training for Text Classification with Noisy Labels

Co-Regularized Adversarial Learning for Multi-Domain Text Classification

Virtual Adversarial Training for Semi-supervised Breast Mass Classification

Reciprocal Adversarial Learning for Brain Tumor Segmentation: A Solution to BraTS Challenge 2021 Segmentation Task

Multilingual and Multilabel Emotion Recognition using Virtual Adversarial Training