SOURCE-LINKED INTELLIGENCE

One Adapter, Many Tasks: Task-Conditioned Feature Transformations for Continual Learning

arXiv · AI, language, vision and robotics · article · Aug 31, 2026 · UTC

Class-incremental learning (CIL) requires a model to incrementally learn tasks that contain new classes without accessing earlier training data while preserving the ability to recognize all seen classes. Recently, pretrained-model-based approaches have become prevalent by adapting a frozen backbone with additional lightweight trainable modules. Existing methods, however, exhibit limitations: task-specific adapters learn explicit per-task representations but are parameter- and computation-inefficient, while LoRA-based merging methods combine per-task LoRA parameters into a single model whose st

Read original source ↗ Open in workspace

recordType

paper

region

Global