利用合成任务先验预训练跨视图的可复用推理

中文摘要 AI 辅助

现代预训练编码器使得来自异构视图的表示越来越具有可复用性，但确定视图效用和组合证据的过程仍需为每个下游任务重新学习。因此，关于视图相关性、互补性、可靠性和缺失性的知识会被反复丢弃，而非在任务间迁移。为此，我们将多视图学习重新表述为学习一个可复用的、任务条件化的推理过程，而非固定的融合函数。基于这一视角，我们提出SIMPLE，这是一种先验适配的多视图上下文学习者，通过对少量带标记的支持集进行条件化来预测查询标签。由于现有真实世界数据集仅覆盖有限范围的视图配置和任务结构，我们在嵌入空间中构建了可控的合成任务先验，该先验生成具有不同类别结构、共享及视图特定因素、表示几何、跨视图依赖、可靠性水平、缺失模式和分布偏移的多样支持-查询 episodes。分层推理架构随后在视图内、跨视图以及支持和查询样本间执行推理。在多视图和多组学基准上的实验表明，SIMPLE的冻结变体无需更新推理骨干即可达到有竞争力的性能，而轻量级适配器校准在大多数评估数据集上取得领先性能。综合来看，冻结、少样本和缺失视图设置下的结果支持核心假设：多视图推理本身可被预训练并复用，而轻量级适配器校准则在需要时提供任务特定的对齐。

英文摘要

Modern pretrained encoders make representations from heterogeneous views increasingly reusable, but the procedure that determines view utility and combines evidence is still relearned for each downstream task. Consequently, knowledge about view relevance, complementarity, reliability, and missingness is repeatedly discarded rather than transferred across tasks. We therefore reformulate multi-view learning as learning a reusable, task-conditioned inference procedure rather than a fixed fusion function. Based on this perspective, we propose SIMPLE, a prior-fitted multi-view in-context learner that predicts query labels by conditioning on a small labeled support set. Since existing real-world datasets cover only a limited range of view configurations and task structures, we construct a controllable synthetic task prior in embedding space. It generates diverse support-query episodes with varying class structures, shared and view-specific factors, representation geometries, cross-view dependencies, reliability levels, missingness patterns, and distribution shifts. A hierarchical inference architecture then performs reasoning within views, across views, and across support and query samples. Experiments on multi-view and multi-omics benchmarks demonstrate that the frozen variant of SIMPLE achieves competitive performance without updating the inference backbone, while lightweight adapter calibration attains leading performance on most evaluated datasets. Together, the results under frozen, one-shot, and missing-view settings support the central hypothesis that multi-view reasoning itself can be pretrained and reused, while lightweight adapter calibration provides task-specific alignment when needed.

发表机构

• Hong Kong Baptist University（香港浸会大学）
• Zhejiang University（浙江大学）

机构由 AI 辅助整理，请以论文原文为准。