Residual Orthogonal Adaptation for Parameter-Efficient Continual Learning
Co-authored
conference Manuscript under peer review at a top-tier international conference 2026
AI Student
Cai Selvas-Sala
ETH Zurich
MSc Computer Science student at ETH Zurich, specialising in Machine Intelligence. Artificial Intelligence graduate from the Universitat Politècnica de Catalunya (UPC), with an exchange semester at the Technical University of Munich (TUM).
First author of a CVPR 2026 main-conference paper on multimodal machine unlearning in CLIP models, with research experience at the Computer Vision Center (CVC). My Bachelor's Thesis focused on continual learning and capability preservation in foundation models.
Interests: foundation models, representation learning, continual and lifelong learning, world models, multimodal learning, computer vision, spatial intelligence, and robot learning.
Research
Selected publications
A quick view of recent work, venues, and project links.
SALMUBench: A Benchmark for Sensitive Association-Level Multimodal Unlearning
Cai Selvas-Sala, Lei Kang, Lluis Gomez
conference Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2026
Updates
Latest news
Recent milestones, announcements, and project activity.
Website launched!
Launched my new academic website built with Lumina.