Consensus-gated Multi-Agent Neural Architecture Search for Seismic Fault Segmentation

2026-08-13

AI Summary

This study addresses the challenges of architectural design and high computational costs associated with traditional Neural Architecture Search (NAS) in few-shot seismic fault segmentation. We propose a Multi-Agent Consensus-Gated NAS system that pioneers code-level architecture search via Large Language Model debates, thereby eliminating predefined operation space constraints. Through multi-agent collaboration and automated verification loops, this approach enables efficient model discovery. Experiments demonstrate that training only eight candidate models—requiring approximately one GPU-day—yields an optimal 425K-parameter network achieving an F1-score of 0.578. This performance significantly surpasses larger architectures such as U-Net, validating the method’s capability to automatically construct high-performance, domain-specific networks under strict computational constraints.

Abstract

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.

Problem

• Seismic Fault Segmentation
• Neural Architecture Search
• Data Scarcity
• Resource-constrained Optimization

Innovation

• Multi-Agent NAS
• Consensus-gated LLM
• Source-code-level Search
• Seismic Fault Segmentation
• Low-resource Architecture Discovery