{"type":"ScholarlyArticle","title":"MOBO: A Merging-Oriented Bi-Level Optimization Framework for Class Incremental Learning","authors":[{"name":"Siyu Zhang","slug":"siyu-zhang","given":"Siyu","family":"Zhang","affiliation":""},{"name":"Wen Wang","slug":"wen-wang","given":"Wen","family":"Wang","affiliation":""},{"name":"Wenju Sun","slug":"wenju-sun","given":"Wenju","family":"Sun","affiliation":""},{"name":"Qingyong Li","slug":"qingyong-li","given":"Qingyong","family":"Li","affiliation":""},{"name":"Yangliao Geng","slug":"yangliao-geng","given":"Yangliao","family":"Geng","affiliation":""}],"abstract":"Class-Incremental Learning (CIL) aims to enable models to sequentially learn new tasks while retaining knowledge from previous ones. Recently, merging-based pre-trained CIL methods have gained significant attention due to their competitive performance and high inference efficiency. However, most existing approaches decouple training from merging, neglecting the compatibility among task-specific adapters. This incompatibility introduces severe conflicts during integration, resulting in catastrophic forgetting and notable performance degradation. To address this limitation, we propose MOBO, a Merging-Oriented Bi-Level Optimization framework that synergizes the optimization of the current task model with the performance of the final merged model. At the upper level, we introduce a global loss that anticipates the merged model's behavior, guiding the current task parameters toward a solution space that facilitates effective merging. At the lower level, we employ a dynamic weighted merging strategy to optimize merging coefficients and update the merged model. By alternately optimizing the task-specific and merged models, MOBO effectively mitigates the performance loss caused by incompatible adapter integration. Comprehensive experiments on CIFAR-100, CUB-200, ImageNet-R, ImageNet-A, and VTAB demonstrate the superiority of our approach, highlighting its robustness and scalability, particularly on long task sequences.","doi":"10.24963/ijcai.2026/593","url":"https://ijcai.org/proceedings/2026/593","pdf":"https://ijcai.org/proceedings/2026/0593.pdf","markdown":"https://ijcai.org/proceedings/2026/593.md","bibtex":"https://ijcai.org/proceedings/2026/bibtex/593","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":"Machine Learning","track":"Main Track","number":593,"pages":{"first":5324,"last":5332},"keywords":["Machine Learning: Classification","Machine Learning: Incremental learning"],"videos":[],"references":[],"license_note":"Copyright International Joint Conferences on Artificial Intelligence."}