实验室研究成果SHAPE被 ICDM 2026 录用 Lab Paper SHAPE Accepted by ICDM 2026
近日，实验室研究成果 SHAPE: Spatial Hyperbolic-Embedding Augmented Pretraining for Estimated Time of Arrival on Obfuscated Trajectories 被ICDM 2026接收。预计到达时间（ETA）是网约车、外卖配送、物流调度等位置服务的核心算法能力，其精度高度依赖高质量、高精度的 GPS 轨迹数据。该论文针对轨迹隐私混淆场景下的ETA预测这一现实难题，提出了一种基于空间双曲嵌入增强预训练的全新方法。
The estimated arrival time (ETA) is a core algorithmic capability for location-based services such as ride-hailing, food delivery, and logistics scheduling. Its accuracy is highly dependent on high-quality and high-precision GPS trajectory data. This paper addresses the practical challenge of predicting the estimated arrival time (ETA) in scenarios where trajectory privacy is obscured and proposes a novel method based on spatial hyperbolic embedding augmented pre-training.