Weizhu Qian

The urban foundation model is critical for trajectory-based mobile applications, which require accurate synthesis of paths that adhere to spatial constraints (road networks) and contextual constraints (e.g., weather, traffic). However, existing methods predominantly rely on task-specific models, which fail to holistically capture and integrate diverse spatial patterns (e.g., connectivity) and temporal dynamics (e.g., periodicity, trends) within a cohesive framework, limiting their generalization across diverse prediction tasks. To bridge this gap, we propose AutoDiff, a diffusion-based model generating trajectories on spatial temporal graph (STG), which establishes a new paradigm for trajectory generation as a foundation model for sequential spatiotemporal data. Specifically, we disentangle complex spatiotemporal features as generalizable segment-wise time slices on road networks through autoregressive diffusion generation, which not only enforces realistic trajectory connectivity within road networks, but also enables knowledge transfer across tasks like trajectory recovery and travel time prediction. Besides, we design a confidence-based early-exiting mechanism to eliminate redundant denoising steps without sacrificing quality, enabling scalable applications in mobility analytics. Extensive experiments on three real-world urban trajectory datasets demonstrate the superior performance of AutoDiff in path prediction, trajectory recovery and time estimation tasks, outperforming task-specific baselines while maintaining computational efficiency.

Cyber-physical system sensors emit multivariate time series (MTS) that monitor physical system processes. Such time series generally capture unknown numbers of states, each with a different duration, that correspond to specific conditions, e.g., "walking" or "running" in human-activity monitoring. Unsupervised identification of such states facilitates storage and processing in subsequent data analyses, as well as enhances result interpretability. Existing state-detection proposals face three challenges. First, they introduce substantial computational overhead, rendering them impractical in resourceconstrained or streaming settings. Second, although state-of-the-art (SOTA) proposals employ contrastive learning for representation, insufficient attention to false negatives hampers model convergence and accuracy. Third, SOTA proposals predominantly only emphasize offline non-streaming deployment, we highlight an urgent need to optimize online streaming scenarios. We propose E2Usd that enables efficient-yet-accurate unsupervised MTS state detection. E2Usd exploits a Fast Fourier Transform-based Time Series Compressor (fftCompress) and a Decomposed Dual-view Embedding Module (ddEM) that together encode input MTSs at low computational overhead. Additionally, we propose a False Negative Cancellation Contrastive Learning method (fnccLearning) to counteract the effects of false negatives and to achieve more cluster-friendly embedding spaces. To reduce computational overhead further in streaming settings, we introduce Adaptive Threshold Detection (adaTD). Comprehensive experiments with six baselines and six datasets offer evidence that E2Usd is capable of SOTA accuracy at significantly reduced computational overhead. Our code is available at https://github.com/AI4CTS/E2Usd.

Traffic speed forecasting has been a very active research area as it is essential for Intelligent Transportation Systems. Although a plethora of deep learning methods have been proposed for traffic speed forecasting, the majority of them can only make point-wise prediction, which may not provide enough information for critical real-world scenarios where prediction confidence also need to be estimated, e.g., route planning for ambulances and rescue vehicles. To address this issue, we propose a novel uncertainty-aware deep learning method coined Uncertainty-Aware Temporal Graph Convolutional Network (UAT-GCN). UAT-GCN employs a Graph Convolutional Network and Gated Recurrent Unit based architecture to capture spatio-temporal dependencies. In addition, UAT-GCN consists of a specialized regressor for estimating both epistemic (model-related) and aleatoric (data-related) uncertainty. In particular, UAT-GCN utilizes Monte Carlo dropout and predictive variances to estimate epistemic and aleatoric uncertainty, respectively. In addition, we also consider the recursive dependency between predictions to further improve the forecasting performance. An extensive empirical study with real datasets offers evidence that the proposed model is capable of advancing current state-of-the-arts in terms of point-wise forecasting and quantifying prediction uncertainty with high reliability. The obtained results suggest that, compared to existing methods, the RMSE and MAE of the proposed model on the SZ-taxi dataset are reduced by $2.15\%$ and $7.23\%$ , respectively; the RMSE and MAE of the proposed model on the Los-loop dataset are reduced by $4.17\%$ and $8.53\%$ , respectively.

Uncertainty is an essential consideration for time series forecasting tasks. In this work, we focus on quantifying the uncertainty of traffic forecasting from a unified perspective. We develop a novel traffic forecasting framework, namely Deep Spatio-Temporal Uncertainty Quantification (DeepSTUQ), which can estimate both aleatoric and epistemic uncertainty. Specifically, we first leverage a spatio-temporal model to model the complex spatio-temporal correlations of traffic data. Subsequently, two independent sub-neural networks maximizing the heterogeneous log-likelihood are developed to estimate aleatoric uncertainty. To estimate epistemic uncertainty, we combine the merits of variational inference and deep ensembling by integrating the Monte Carlo dropout and the Adaptive Weight Averaging re-training methods, respectively. Furthermore, to relax the Gaussianity assumption, mitigate overfitting, and improve horizon-wise uncertainty quantification performance, we define a new calibration method called Multi-horizon Conformal Calibration (MHCC). Finally, we provide a theoretical analysis of the proposed unified approach based on the PAC-Bayes theory. Extensive experiments are conducted on four public datasets, and the empirical results suggest that the proposed method outperforms state-of-the-art methods in terms of both point prediction and uncertainty quantification.

Spatial Crowdsourcing (SC) is finding widespread application in today's online world. As we have transitioned from desktop crowdsourcing applications (e.g., Wikipedia) to SC applications (e.g., Uber), there is a sense that SC systems must not only provide effective task assignment but also need to ensure privacy. To achieve these often-conflicting objectives, we propose a framework, Task Assignment with Federated Preference Learning, that performs task assignment based on worker preferences while keeping the data decentralized and private in each platform center (e.g., each delivery center of an SC company). The framework includes a federated preference learning phase and a task assignment phase. Specifically, in the first phase, we build a local preference model for each platform center based on historical data. We provide means of horizontal federated learning that makes it possible to collaboratively train these local preference models under the orchestration of a central server. Specifically, we provide a practical method that accelerates federated preference learning based on stochastic controlled averaging and achieves low communication costs while considering data heterogeneity among clients. The task assignment phase aims to achieve effective and efficient task assignment by considering workers' preferences. Extensive evaluations on real data offer insight into the effectiveness and efficiency of the paper's proposals.

Predicting smartphone users location with WiFi fingerprints has been a popular research topic recently. In this work, we propose two novel deep learning-based models, the convolutional mixture density recurrent neural network and the VAE-based semi-supervised learning model. The convolutional mixture density recurrent neural network is designed for path prediction, in which the advantages of convolutional neural networks, recurrent neural networks and mixture density networks are combined. Further, since most of real-world datasets are not labeled, we devise the VAE-based model for the semi-supervised learning tasks. In order to test the proposed models, we conduct the validation experiments on the real-world datasets. The final results verify the effectiveness of our approaches and show the superiority over other existing methods.

Multi-task learning (MTL) is an important subject in machine learning and artificial intelligence. Its applications to computer vision, signal processing, and speech recognition are ubiquitous. Although this subject has attracted considerable attention recently, the performance and robustness of the existing models to different tasks have not been well balanced. This article proposes an MTL model based on the architecture of the variational information bottleneck (VIB), which can provide a more effective latent representation of the input features for the downstream tasks. Extensive observations on three public data sets under adversarial attacks show that the proposed model is competitive to the state-of-the-art algorithms concerning the prediction accuracy. Experimental results suggest that combining the VIB and the task-dependent uncertainties is a very effective way to abstract valid information from the input features for accomplishing multiple tasks.

Analyzing human mobility with geo-location data collected from smartphones has been a hot research topic in recent years. In this paper, we attempt to discover daily mobile patterns using the GPS data. In particular, we view this problem from a probabilistic perspective. A non-parametric Bayesian modeling method, the Infinite Gaussian Mixture Model (IGMM) is used to estimate the probability density of the daily mobility. We also utilize the Kullback-Leibler (KL) divergence as the metrics to measure the similarity of different probability distributions. Combining the IGMM and the KL divergence, we propose an automatic clustering algorithm to discover mobility patterns for each individual user. Finally, the effectiveness of our method is validated on the real user data collected from different real users.