用锚点对齐多模型金融文本表示，提升预测稳定性。
FinAnchor: Aligned Multi-Model Representations for Financial Prediction
- 选一个基准嵌入空间，线性映射对齐其他模型表示
- 在多个金融任务上优于单模型和传统集成方法
- 无需微调大模型，适合不同时间周期的金融预测
从长篇金融文档中进行预测面临巨大挑战，因为有效信号稀疏且被噪声掩盖，且生成嵌入的最优大语言模型随任务和时间变化。本文提出FinAnchor（金融锚定表示），一种轻量级框架，无需微调底层模型即可融合多个大语言模型的嵌入。FinAnchor通过选择一个锚定嵌入空间，并学习线性映射将其他模型的表示对齐到该空间，再聚合形成统一表示用于下游预测。在多个金融自然语言处理任务中，FinAnchor始终优于强单模型基线和标准集成方法，证明了锚定异构表示在鲁棒金融预测中的有效性。
原文摘要 · Abstract (English)
Financial prediction from long documents involves significant challenges, as actionable signals are often sparse and obscured by noise, and the optimal LLM for generating embeddings varies across tasks and time periods. In this paper, we propose FinAnchor(Financial Anchored Representations), a lightweight framework that integrates embeddings from multiple LLMs without fine-tuning the underlying models. FinAnchor addresses the incompatibility of feature spaces by selecting an anchor embedding space and learning linear mappings to align representations from other models into this anchor. These aligned features are then aggregated to form a unified representation for downstream prediction. Across multiple financial NLP tasks, FinAnchor consistently outperforms strong single-model baselines and standard ensemble methods, demonstrating the effectiveness of anchoring heterogeneous representations for robust financial prediction.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。