标题： 思维链增强的浅层变压器用于无线符号检测 显示英文标题

标题： Chain-of-Thought Enhanced Shallow Transformers for Wireless Symbol Detection

摘要： Transformer在解决无线通信问题方面显示出潜力，特别是通过上下文学习（ICL），其中模型通过提示适应新任务而无需更新模型。 然而，之前的基于ICL的Transformer模型依赖于深度架构，具有许多层以实现令人满意的性能，导致存储和计算成本显著增加。 在本工作中，我们提出了CHain Of thOught Symbol dEtection（CHOOSE），一种增强思维链的浅层Transformer框架，用于无线符号检测。 通过在隐藏空间中引入自回归潜在推理步骤，CHOOSE在不增加模型深度的情况下显著提高了浅层模型（1-2层）的推理能力。 这种设计使轻量级Transformer能够实现与更深层模型相当的检测性能，使其非常适合在资源受限的移动设备上部署。 实验结果表明，我们的方法优于传统的浅层Transformer，并且实现了与深层Transformer相当的性能，同时保持了存储和计算效率。 这为在计算资源有限的无线接收器中实现基于Transformer的算法提供了一个有前景的方向。 显示英文摘要

摘要： Transformers have shown potential in solving wireless communication problems, particularly via in-context learning (ICL), where models adapt to new tasks through prompts without requiring model updates. However, prior ICL-based Transformer models rely on deep architectures with many layers to achieve satisfactory performance, resulting in substantial storage and computational costs. In this work, we propose CHain Of thOught Symbol dEtection (CHOOSE), a CoT-enhanced shallow Transformer framework for wireless symbol detection. By introducing autoregressive latent reasoning steps within the hidden space, CHOOSE significantly improves the reasoning capacity of shallow models (1-2 layers) without increasing model depth. This design enables lightweight Transformers to achieve detection performance comparable to much deeper models, making them well-suited for deployment on resource-constrained mobile devices. Experimental results demonstrate that our approach outperforms conventional shallow Transformers and achieves performance comparable to that of deep Transformers, while maintaining storage and computational efficiency. This represents a promising direction for implementing Transformer-based algorithms in wireless receivers with limited computational resources.