Task-Anchored Representation Shaping for Pre-Trained Model-Based Continual Learning
2026-08-17 • Machine Learning
Machine Learning
AI summaryⓘ
The authors introduce TAILS, a small add-on module designed to improve how pre-trained models handle learning multiple tasks over time. While pre-trained models help adapt to new tasks, they struggle to correctly identify which task an input belongs to when tasks overlap. TAILS fixes this by using fixed task references to adjust the input features directly, improving task recognition and classification without changing the original model or its classifiers. Their tests show TAILS boosts performance with little extra cost.
pre-trained modelscontinual learningtask inferencefeature representationlatent recallclassificationcross-task ambiguitypost-training moduletask anchorsparameter overhead
Authors
Zhiming Xu, Huiyu Yi, Zhen-Hao Xie, Baile Xu, Furao Shen, Jian Zhao, Suorong Yang
Abstract
Pre-trained models (PTMs) provide a strong foundation for continual learning by offering stable representations that facilitate lightweight adaptation to new tasks. However, adapting well to each task does not ensure reliable inference over all learned tasks. Since task boundaries are often artificial and semantically entangled, an input from an unknown task can remain ambiguous even with strong PTM features, making cross-task prediction a key bottleneck. We propose Task-Anchored Inference Latent Shaping (TAILS), a lightweight post-PTM module that can be integrated into diverse continual learners and optimized through a decoupled step. TAILS uses fixed task anchors as persistent references to accumulated knowledge. It interprets each sample's feature representation relative to these references, then composes relevant evidence across tasks into latent recall. Rather than selecting a task-specific path or adjusting classifier outputs, TAILS uses latent recall to directly correct the feature representation before prediction. It therefore resolves cross-task ambiguity at the representation level, while leaving the original PTM, method-specific modules, and classifier unchanged. Extensive experiments across multiple PTM-based continual learning paradigms show that TAILS can improve classification and task-inference performance with modest parameter overhead and negligible inference cost.