{"type":"ScholarlyArticle","title":"CE-VFAL: A Novel Framework for Communication-Efficient Vertical Federated Adversarial Learning","authors":[{"name":"Tianxing Man","slug":"tianxing-man","given":"Tianxing","family":"Man","affiliation":""},{"name":"Jinjie Fang","slug":"jinjie-fang","given":"Jinjie","family":"Fang","affiliation":""},{"name":"Ganyu Wang","slug":"ganyu-wang","given":"Ganyu","family":"Wang","affiliation":""},{"name":"Yu Bai","slug":"yu-bai","given":"Yu","family":"Bai","affiliation":""},{"name":"Zhaogeng Liu","slug":"zhaogeng-liu","given":"Zhaogeng","family":"Liu","affiliation":""},{"name":"Bin Gu","slug":"bin-gu","given":"Bin","family":"Gu","affiliation":""},{"name":"Yi Chang","slug":"yi-chang","given":"Yi","family":"Chang","affiliation":""}],"abstract":"Vertical Federated Learning (VFL) involves multiple participants collaborating to train machine learning models on distinct feature sets from the same data samples. \r\nThis training paradigm with distributed updating focuses on secure and efficient communication.\r\nNevertheless, the trained models exhibit heightened vulnerability to adversarial attacks during inference, which can provoke misclassification.\r\nAdversarial Training (AT), which involves exposing models to intentionally crafted misleading examples during training, is widely regarded as the most effective method for enhancing model robustness. \r\nHowever, the significant communication costs entailing such example generation within the VFL context pose an open challenge to developing a Vertical Federated Adversarial Learning (VFAL) framework.\r\nTo this end, we introduce a Communication-Efficient Vertical Federated Adversarial Learning framework, named CE-VFAL. \r\nCE-VFAL framework incorporates the lazy propagation principle, confining most propagations to client models during adversarial updates, thereby minimizing frequent client-server interactions. \r\nMoreover, CE-VFAL seamlessly integrates Zeroth Order Optimization (ZOO) into communication, effectively reducing communication load by transmitting the loss difference derived from the raw and perturbed embeddings for multiple point estimation.\r\nFurthermore, our theoretical analysis demonstrates the sublinear convergence rate by containing the errors caused by multi-source approximate gradients.\r\nExtensive experiments corroborate the robust performance while significantly reducing communication costs.","doi":"10.24963/ijcai.2026/79","url":"https://ijcai.org/proceedings/2026/79","pdf":"https://ijcai.org/proceedings/2026/0079.pdf","markdown":"https://ijcai.org/proceedings/2026/79.md","bibtex":"https://ijcai.org/proceedings/2026/bibtex/79","volume":{"slug":"2026","title":"Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence","short_title":"IJCAI 2026","url":"https://ijcai.org/proceedings/2026/","year":2026,"location":"Bremen, Germany","dates":"15-21 August 2026","editors":"Diego Calvanese","isbn":"978-1-956792-09-6","publisher":"International Joint Conferences on Artificial Intelligence Organization","doi_prefix":"10.24963/ijcai.2026"},"section":"AI Ethics, Trust, Fairnes","track":"Main Track","number":79,"pages":{"first":697,"last":705},"keywords":["Machine Learning: Federated learning","AI Ethics, Trust, Fairnes: Safety and robustness"],"videos":[],"references":[],"license_note":"Copyright International Joint Conferences on Artificial Intelligence."}