发表机构
Information Technology University (ITU); King Fahd University of Petroleum and Minerals（信息技术大学（ITU）; 法赫德国王石油与矿业大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
提出共识门控多智能体NAS系统，利用三个LLM达成共识搜索架构，发现参数42.5万的\textit{ours}模型，在Thebe断层数据集上性能优于多个基准模型，成本低、效率高。
AI 中文摘要
地震断层分割所用的神经网络常借鉴计算机视觉与医学成像领域的模型，这类模型通常在规模大得多的标注数据上训练。而地球物理应用中常见的标注数据预算有限，在这种条件下优化网络架构并非易事。手动设计数据最优架构耗时，传统神经架构搜索（NAS）受限于人工定义的搜索空间与大量计算预算。本文提出一种智能体NAS系统：由Claude、GPT-5.1和Gemini~2.5~Pro三个大语言模型组成的小组，针对每个候选架构展开辩论直至达成一致共识，编写完整的PyTorch实现代码，交叉审核后提交至自动化的验证-训练-评分循环，该循环设置了严格的45万参数预算、保留/回退机制以及失败机制的记忆功能。该系统基于源代码而非预定义的操作菜单运行，搜索在单个消费级GPU上完成，仅训练了8个候选模型。最终发现了\textit{ours}模型：一款参数规模为42.5万的编解码器，带有条带池化瓶颈、挤压激励门控、非对称单卷积解码器以及特征金字塔融合颈部。在与所有基准模型相同的协议下，基于Thebe断层数据集的切片进行训练，该模型达到了所有测试模型中最高的F1值（0.578）和IoU值，同时是参数最小的模型，性能优于已发表的同容量U-Net（3100万参数，F1值0.484）、DeepLabV3-ResNet50（3960万参数，F1值0.516）以及Attention U-Net（183万参数，F1值0.502）。此次搜索共进行了101次LLM调用（约115万输入token/39万输出token），耗时约1个GPU天，表明共识门控LLM小组是一种实用、低成本的领域特定架构发现途径。
英文摘要
Neural networks for seismic fault segmentation are often borrowed from computer vision and medical imaging domains where they train under relatively much larger labeled data resources. Optimizing their architecture under tight labeled data budgets as are common in geophysical applications is not a trivial problem. Manually designing data-optimal architectures is time-consuming while classical neural architecture search (NAS) is restricted to hand-crafted search spaces and large compute budgets. We present an agentic NAS system in which a panel of three large language models (Claude, GPT-5.1, and Gemini~2.5~Pro) debates each candidate architecture to unanimous consensus, authors the complete PyTorch implementation, cross-reviews it, and submits it to an automated validate-train-score loop with a hard 450K parameter budget, keep-or-revert lineage, and a memory of failed mechanisms. Operating on source code rather than a predefined operation menu, the search ran on a single consumer GPU and trained only eight candidates. It discovered \ours{}: a 425K-parameter encoder-decoder with a strip-pooling bottleneck, squeeze-and-excitation gating, an asymmetric one-conv decoder, and a feature-pyramid fusion neck. Trained under a protocol identical to all baselines on sections derived from the Thebe fault dataset, it attains the highest F1 (0.578) and IoU of all models tested while being the smallest, outperforming a published-capacity U-Net (31M parameters, F1 0.484), DeepLabV3-ResNet50 (39.6M, 0.516), an Attention U-Net(1.83M, 0.502). The search cost 101 LLM calls ($\sim$1.15M input / 0.39M output tokens) and roughly one GPU-day, making consensus-gated LLM panels a practical, low-cost route to domain-specific architecture discovery.
