Scaling Cheap Synthetic Pretraining Improves Indoor Radio-Map Prediction

查看原文 / View Original 编辑 / Edit

Computer Science > Machine Learning arXiv:2610.19015 (cs) [Submitted on 13 Oct 2026] Title: Scaling Cheap Synthetic Pretraining Improves Indoor Radio-Map Prediction Authors: Khoren Petrosyan, Artashes Mkrtchyan, Rafayel Mkrtchyan, Hrant Khachatrian, Theofanis P. Raptis.

计算机科学 > 机器学习 arXiv:2610.19015 (cs) [提交于 2026 年 10 月 13 日] 标题：扩展廉价合成预训练以改进室内无线电地图预测作者：Khoren Petrosyan, Artashes Mkrtchyan, Rafayel Mkrtchyan, Hrant Khachatrian, Theofanis P. Raptis。

Abstract: Real measurements are too limited to train an accurate indoor pathloss predictor that works across a wide range of buildings. Large-scale synthetic pretraining offers a way forward, but real building layouts are also scarce, and accurate ray-tracing simulators are slow, expensive, and often proprietary.

摘要：真实的测量数据过于有限，无法训练出适用于各种建筑的精确室内路径损耗预测器。大规模合成预训练提供了一种前进的方向，但真实的建筑布局同样稀缺，且精确的射线追踪模拟器速度慢、成本高，且通常是专有的。

We address both problems by introducing a generator that automatically creates diverse indoor layouts and a fast, open-source approximate ray tracer that computes radio maps for them. Using this pipeline, we generate 128M synthetic samples, pretrain ResNet-101 and Swin-T encoder-decoders, and fine-tune them on high-quality benchmark data.

我们通过引入一个自动创建多样化室内布局的生成器，以及一个用于计算无线电地图的快速开源近似射线追踪器，解决了这两个问题。利用这一流程，我们生成了 1.28 亿个合成样本，预训练了 ResNet-101 和 Swin-T 编码器-解码器，并在高质量基准数据上对它们进行了微调。

Despite simulator mismatch, our ResNet-101 model reduces error by 15% on average relative to the best known methods across five indoor radio-map prediction tasks and improves transfer on a small real-measurement dataset. Both architectures show predictable data-scaling trends: power laws fitted on 2M-16M samples predict the mean normalized downstream RMSE at 32M-128M with 3.33% mean absolute percentage error for ResNet-101 and 7.10% for Swin-T.

尽管存在模拟器不匹配的问题，我们的 ResNet-101 模型在五个室内无线电地图预测任务中，相较于目前已知最好的方法，平均误差降低了 15%，并改善了在小型真实测量数据集上的迁移效果。两种架构都表现出可预测的数据扩展趋势：在 200 万至 1600 万样本上拟合的幂律，能够以 3.33%（ResNet-101）和 7.10%（Swin-T）的平均绝对百分比误差，预测 3200 万至 1.28 亿样本规模下的平均归一化下游 RMSE。

These results show that pretraining on large amounts of cheap simulated data can improve indoor radio-map prediction, even when real measurements and high-quality simulations are limited.

这些结果表明，即使在真实测量数据和高质量模拟数据有限的情况下，利用大量廉价模拟数据进行预训练也能改善室内无线电地图的预测效果。