MoCo-EA: Exploiting Adversarial Mode Connectivity for Efficient Evolutionary Attacks
Not on the map. Why?
Abstract (arXiv)
Evolutionary algorithms for adversarial attacks leverage population-based search to discover perturbations without gradient information, but suffer from inefficient crossover operations that destroy adversarial properties through discrete interpolation. We introduce Mode Connectivity Evolutionary Attack (MoCo-EA), which replaces traditional crossover with a novel B'ezier crossover operator that optimizes perturbations along a continuous B'ezier curve between parent perturbations. Our key insight is that adversarial examples lie on connected manifolds where intermediate points maintain and often enhance attack effectiveness. We demonstrate three findings: (1) Successful adversarial perturbations exhibit mode connectivity; (2) Intermediate points along optimized paths achieve higher transferability than endpoints; (3) B'ezier crossover dramatically outperforms discrete genetic operations while reducing convergence time and query requirements. By exploiting the geometric structure of adversarial space through path optimization, MoCo-EA provides an efficient and reliable method. Our work challenges the traditional view of adversarial examples as isolated points and opens new directions for both attack generation and defense research.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 28 of its 35 references that are in the map. First posted on arXiv 18 May 2026.
Builds on
28 of 35 references.
References · 28
- 2015Deep Residual Learning for Image Recognitioncited by 52,096
- 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
- 2014Explaining and Harnessing Adversarial Examplescited by 7,364
- 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,415
- 2016Towards Evaluating the Robustness of Neural Networkscited by 2,998
- 2020Reliable evaluation of adversarial robustness with an ensemble of diverse parameter-free attackscited by 1,451
- 2018Obfuscated Gradients Give a False Sense of Security: Circumventing Defenses to Adversarial Examplescited by 1,268
- 2017ZOO: Zeroth Order Optimization based Black-box Attacks to Deep Neural Networks without Training Substitute Modelscited by 666
- 2018Improving Transferability of Adversarial Examples with Input Diversitycited by 599
- 2018Loss Surfaces, Mode Connectivity, and Fast Ensembling of DNNscited by 588
- 2017One pixel attack for fooling deep neural networkscited by 582
- 2017Decision-Based Adversarial Attacks: Reliable Attacks Against Black-Box Machine Learning Modelscited by 462
16 more
- 2019Evading Defenses to Transferable Adversarial Examples by Translation-Invariant Attackscited by 442
- 2019HopSkipJumpAttack: A Query-Efficient Decision-Based Attackcited by 333
- 2018Essentially No Barriers in Neural Network Energy Landscapecited by 330
- 2016Topology and Geometry of Half-Rectified Network Optimizationcited by 102
- 2018GenAttack: Practical Black-box Attacks with Gradient-Free Optimizationcited by 82
- 2022Boosting the Transferability of Adversarial Attacks with Reverse Adversarial Perturbationcited by 68
- 2023How Deep Learning Sees the World: A Survey on Adversarial Attacks & Defensescited by 66
- 2022Practical Evaluation of Adversarial Robustness via Adaptive Auto Attackcited by 44
- 2021On Fast Adversarial Robustness Adaptation in Model-Agnostic Meta-Learningcited by 29
- 2019POBA-GA: Perturbation Optimized Black-Box Adversarial Attacks via Genetic Algorithmcited by 14
- 2021RAILS: A Robust Adversarial Immune-inspired Learning Systemcited by 8
- 2025Optimizing Robustness and Accuracy in Mixture of Experts: A Dual-Model Approachcited by 8
- 2021ASK: Adversarial Soft k-Nearest Neighbor Attack and Defensecited by 4
- 2024Input Space Mode Connectivity in Deep Neural Networkscited by 4
- 2025Exploring Nonlinear Pathway in Parameter Space for Machine Unlearningcited by 4
- 2024Deep Adversarial Defense Against Multilevel-Lp Attackscited by 1
Built on by
Citing papers per year: 2026: 1.
1 paper.