Adversarial Uncertainty

Adversarial uncertainty focuses on understanding and mitigating the uncertainty inherent in models' predictions when faced with adversarial inputs—data designed to fool the system. Current research explores how to generate less certain adversarial examples to improve model robustness and generalization, often employing deep learning techniques and incorporating concepts like differential privacy to enhance model security. This work is crucial for building more trustworthy and reliable AI systems, particularly in applications where security and robustness are paramount, such as cybersecurity and medical diagnosis. The development of model-agnostic methods for quantifying and addressing both data and model uncertainty is a key area of advancement.

Papers

February 16, 2024

Connect the dots: Dataset Condensation, Differential Privacy, and Adversarial Uncertainty
Kenneth Odoh
Differential Privacy Dataset Condensation Dot Product Adversarial Uncertainty

October 6, 2023

Generating Less Certain Adversarial Examples Improves Robust Generalization
Minxing Zhang, Michael Backes, Xiao Zhang
Adversarial Example Adversarial Training Robust Generalization Adversarial Uncertainty

March 29, 2023

ALUM: Adversarial Data Uncertainty Modeling from Latent Model Uncertainty Compensation
Wei Wei, Jiahuan Zhou, Hongze Li, Ying Wu
Model Uncertainty Uncertainty Aware Data Uncertainty Uncertainty Modeling Adversarial Uncertainty

February 3, 2023

Deep Reinforcement Learning for Cyber System Defense under Dynamic Adversarial Uncertainties
Ashutosh Dutta, Samrat Chatterjee, Arnab Bhattacharya, Mahantesh Halappanavar
Deep Reinforcement Learning Data Driven Defense Algorithm Cyber Defense Autonomous Cyber Adversarial Uncertainty Proactive Cyber Defense

May 1, 2022

Molecular Identification from AFM images using the IUPAC Nomenclature and Attribute Multimodal Recurrent Neural Networks
Jaime Carracedo-Cosme, Carlos Romero-Muñiz, Pablo Pou, Rubén Pérez
Known Molecule Chemical Data Atomic Force Microscopy Adversarial Uncertainty

Adversarial Uncertainty

Papers

Connect the dots: Dataset Condensation, Differential Privacy, and Adversarial Uncertainty

Generating Less Certain Adversarial Examples Improves Robust Generalization

ALUM: Adversarial Data Uncertainty Modeling from Latent Model Uncertainty Compensation

Deep Reinforcement Learning for Cyber System Defense under Dynamic Adversarial Uncertainties

Molecular Identification from AFM images using the IUPAC Nomenclature and Attribute Multimodal Recurrent Neural Networks