arXiv：cs.LG· Wei Hao, Bin Chong, Ronghua Ji, Chen Hou·· 13 小时前
用户轨迹预测统一全局与局部时序信息（论文已撤回）
User Trajectory Prediction Unifying Global and Local Temporal Information
AI 导读
一篇基于 MLP、多尺度卷积神经网络（MSCNN）与 cross-attention 的用户轨迹预测论文被作者 Wei Hao 撤回，未通过同行评审。该方法用 MLP 提取全局时序信息、MSCNN 以多种卷积核尺寸提取局部时序信息，再由 cross-attention 融合二者；实验称在 12 步预测中相比 ModernTCN 将 MSE 降低 5.04%、MAE 降低 4.35%，推理时间相近。
正文
This paper has been withdrawn by Wei Hao
No PDF available, click to view other formats
Abstract:Trajectory prediction is essential for formulating proactive strategies that anticipate user mobility and support advance preparation. Therefore, how to reduce the forecasting error in user trajectory prediction within an acceptable inference time arises as an interesting issue. However, trajectory data contains both global and local temporal information, complicating the extraction of the complete temporal pattern. Moreover, user behavior occurs over different time scales, increasing the difficulty of capturing behavioral patterns. To address these challenges, a trajectory prediction model based on multilayer perceptron (MLP), multi-scale convolutional neural network (MSCNN), and cross-attention (CA) is proposed. Specifically, MLP is used to extract the global temporal information of each feature. In parallel, MSCNN is employed to extract the local temporal information by modeling interactions among features within a local temporal range. Convolutional kernels with different sizes are used in MSCNN to capture temporal information at multiple resolutions, enhancing the model's adaptability to different behavioral patterns. Finally, CA is applied to fuse the global and local temporal information. Experimental results show that our model reduces mean squared error (MSE) by 5.04% and mean absolute error (MAE) by 4.35% compared with ModernTCN in 12-step prediction, while maintaining similar inference time.
| Comments: | This manuscript has not passed peer review |
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2508.02161 [cs.LG] |
| (or arXiv:2508.02161v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2508.02161 arXiv-issued DOI via DataCite |
Submission history
From: Wei Hao [view email]
[v1]
Mon, 4 Aug 2025 08:03:01 UTC (2,766 KB)
[v2]
Fri, 9 Oct 2026 14:26:25 UTC (1 KB) (withdrawn)
来源：arXiv：cs.LG · arxiv.org