Adversarial Optimization

Adversarial optimization focuses on designing robust systems by pitting two competing models against each other, improving the resilience of one (e.g., a language model or classifier) against attacks from the other. Current research emphasizes applications in enhancing the security and privacy of large language models, improving the stability of generative adversarial networks and other minimax optimization problems, and developing efficient methods for resource-constrained environments like federated learning. These advancements are crucial for building reliable and trustworthy AI systems, addressing concerns about adversarial attacks and ensuring fairness in machine learning applications.

Papers

November 6, 2024

ROBIN: Robust and Invisible Watermarks for Diffusion Models with Adversarial Optimization
Huayang Huang, Yu Wu, Qian Wang
Diffusion Model Agnostic Watermarking Robust Watermark Invisible Watermark Adversarial Optimization

August 5, 2024

SEAS: Self-Evolving Adversarial Safety Optimization for Large Language Models
Muxi Diao, Rumei Li, Shiyang Liu, Guogang Liao, Jingang Wang, Xunliang Cai, Weiran Xu
Large Language Model Adversarial Attack Adversarial Prompt Upper Ocean Adversarial Machine Learning Adversarial Optimization

April 7, 2024

Initial Exploration of Zero-Shot Privacy Utility Tradeoffs in Tabular Data Using GPT-4
Bishwas Mandal, George Amariucai, Shuangqing Wei
Large Language Model GPT 4 Tabular Data Task Utility Privacy Utility Adversarial Method Privacy Utility Trade Adversarial Optimization

December 2, 2023

Convergences for Minimax Optimization Problems over Infinite-Dimensional Spaces Towards Stability in Adversarial Training
Takashi Furuya, Satoshi Okuda, Kazuma Suetake, Yoshihide Sawada
Generative Adversarial Network Adversarial Training Gradient Descent Core Stability Minimax Optimization Minimax Optimization Problem Infinite Dimensional Adversarial Optimization

November 1, 2022

Adversarial Training with Complementary Labels: On the Benefit of Gradually Informative Attacks
Jianan Zhou, Jianing Zhu, Jingfeng Zhang, Tongliang Liu, Gang Niu, Bo Han, Masashi Sugiyama
Adversarial Example Adversarial Training Complementary Benefit Complementary Label Label Attack Adversarial Optimization

September 21, 2022

NashAE: Disentangling Representations through Adversarial Covariance Minimization
Eric Yeats, Frank Liu, David Womble, Hai Li
Latent Representation Meaningful Representation Feature Disentanglement Disentanglement Method Disentanglement Metric Concept Disentanglement Adversarial Optimization

September 8, 2022

FADE: Enabling Federated Adversarial Training on Heterogeneous Resource-Constrained Edge Devices
Minxue Tang, Jianyi Zhang, Mingyuan Ma, Louis DiValentin, Aolin Ding, Amin Hassanzadeh, Hai Li, Yiran Chen
Adversarial Training Adversarial Robustness Federated Prompt Cooperation Federated Adversarial Adversarial Optimization

December 15, 2021

Probabilistic Forecasting with Generative Networks via Scoring Rule Minimization
Lorenzo Pacchiardi, Rilwan Adewoyin, Peter Dueben, Ritabrata Dutta
Probabilistic Forecast Generative Network Generative Neural Network Adversarial Optimization

Adversarial Optimization

Papers

ROBIN: Robust and Invisible Watermarks for Diffusion Models with Adversarial Optimization

SEAS: Self-Evolving Adversarial Safety Optimization for Large Language Models

Initial Exploration of Zero-Shot Privacy Utility Tradeoffs in Tabular Data Using GPT-4

Convergences for Minimax Optimization Problems over Infinite-Dimensional Spaces Towards Stability in Adversarial Training

Adversarial Training with Complementary Labels: On the Benefit of Gradually Informative Attacks

NashAE: Disentangling Representations through Adversarial Covariance Minimization

FADE: Enabling Federated Adversarial Training on Heterogeneous Resource-Constrained Edge Devices

Probabilistic Forecasting with Generative Networks via Scoring Rule Minimization