Tuning for Two Adversaries: Enhancing the Robustness Against Transfer and Query-Based Attacks using Hyperparameter Tuning

2025-11-17

AI Summary

This work addresses the challenge of simultaneously enhancing robustness against transfer-based and query-based adversarial attacks. We systematically investigate the impact mechanisms of key hyperparameters—including learning rate, weight decay, momentum, and batch size—on dual robustness. Our study reveals, for the first time, a pronounced antagonistic effect of learning rate: higher values improve query-attack robustness but degrade transfer robustness, and vice versa. Leveraging this insight, we propose a collaborative hyperparameter optimization strategy tailored for joint defense, validated through centralized training, model ensembling, and distributed architectures. Experiments across diverse scenarios demonstrate substantial improvements in dual robustness: up to 64% gain in transfer-attack defense and up to 28% in query-attack defense. Notably, the distributed training architecture achieves the optimal trade-off between the two robustness objectives. This work fills a critical gap in hyperparameter-aware defense design for concurrent transfer and query attacks.

Technology Category

• Machine Learning: Adversarial Learning & Robustness
• Computer Vision: Adversarial Attacks & Robustness
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

In this paper, we present the first detailed analysis of how optimization hyperparameters -- such as learning rate, weight decay, momentum, and batch size -- influence robustness against both transfer-based and query-based attacks. Supported by theory and experiments, our study spans a variety of practical deployment settings, including centralized training, ensemble learning, and distributed training. We uncover a striking dichotomy: for transfer-based attacks, decreasing the learning rate significantly enhances robustness by up to $64%$. In contrast, for query-based attacks, increasing the learning rate consistently leads to improved robustness by up to $28%$ across various settings and data distributions. Leveraging these findings, we explore -- for the first time -- the optimization hyperparameter design space to jointly enhance robustness against both transfer-based and query-based attacks. Our results reveal that distributed models benefit the most from hyperparameter tuning, achieving a remarkable tradeoff by simultaneously mitigating both attack types more effectively than other training setups.

Problem

• Optimizing hyperparameters to enhance robustness against transfer and query attacks
• Analyzing how learning rate affects adversarial robustness in different attack scenarios
• Exploring hyperparameter tuning for simultaneous defense against dual attack types

Innovation

• Hyperparameter tuning enhances robustness against transfer attacks
• Increasing learning rate improves defense against query-based attacks
• Distributed models benefit most from hyperparameter optimization tradeoff