[Time Series Large Language Models for Traffic Flow Forecasting: A Survey and Beyond]

Qingqing Long∗, Shuai Liu∗, Ning Cao∗, Xiao Luo, Wei Ju, Chen Fang, Zheng Fang, Chong Chen, Xiansheng Hua and Yuanchun Zhou†

We made a systematic survey about traffic LLMs.

[Disentangling Dynamics: Advanced, Scalable and Explainable Imputation for Multivariate Time Series]

Shuai Liu, Xiucheng Li, Yile Chen, Yue Jiang, Gao Cong

We propose TIDER and its enhanced version, AdaTIDER. TIDER employs low-rank matrix factorization and disentangled temporal representation to model intricate dynamics. By accounting for various contributing factors such as trend, seasonality, and local bias, TIDER achieves enhanced imputation accuracy. The use of disentangled representations improves the reliability and interpretability of the imputation process. Notably, TIDER demonstrates superior scalability in handling lengthy time series without compromising performance. Furthermore, building upon TIDER, AdaTIDER adaptively learns cross-channel correlations and multi-periods seasonality matrices, surpassing TIDER in terms of imputation accuracy and interpretability while main- taining scalability.

[SAGDFN: A Scalable Adaptive Graph Diffusion Forecasting Network for Multivariate Time Series Forecasting]

Yue Jiang, Xiucheng Li, Yile Chen, Shuai Liu, Weilong Kong, Antonis F. Lentzakis, Gao Cong

We present a Scaleable Adaptive Graph Diffusion Forecasting Network (SAGDFN) to capture complex spatial-temporal correlation for large- scale multivariate time series and thereby, leading to exceptional performance in multivariate time series forecasting tasks. The proposed SAGDFN is scalable to datasets of thousands of nodes without the need of prior knowledge of spatial correlation.

[Multivariate Time-series imputation with disentangled representations]

SHUAI LIU, XIUCHENG LI, GAO CONG, YILE CHEN, YUE JIANG

We propose TIDER, a novel matrix factorization-based method with disentangled temporal representations that account for multiple factors, namely trend, seasonality, and local bias, to model complex dynamics. The learned disen- tanglement makes the imputation process more reliable and offers explainability for imputation results.

[Real-time Transportation Prediction Correction using Reconstruction Error in Deep Learning, TKDD 2020]

SHUAI LIU, GUOJIE SONG, WENHAO HUANG

We propose traffic prediction correction strategy using the reconstruction error in the deep neural network. The reconstruction error can reflect the model’s ability on feature representation and then determine the fitness of an input data to the model. We propose two mechanisms of real-time prediction correction using this reconstruction error: the data driven approach and the model-driven approach.

[高科西路浦东南路口交通优化设计, 2016 上海市青少年人文社会科学论文大赛一等奖 ]

刘帅, 谢欣

论文采样了上海浦东新区高科西路浦东南路路口的实地车辆通行数据，通过计算分析道路平面立交道口通行能力影响系数，调整路口红绿灯时长以及路口车道分布情况。论文的研究结果已经被成功应用于上述路口。