Sensory Temporal Action Detection (STAD) aims to localize and classify human actions within long, untrimmed sequences captured by non-visual sensors such as WiFi or inertial measurement units (IMUs). Unlike video-based TAD, STAD poses unique challenges due to the low-dimensional, noisy, and heterogeneous nature of sensory data, as well as the real-time and resource constraints on edge devices. While recent STAD models have improved detection performance, their high computational cost hampers practical deployment. In this paper, we propose SlimSTAD, a simple yet effective framework that achieves both high accuracy and low latency for STAD. SlimSTAD features a novel Decoupled Channel Modeling (DCM) encoder, which preserves modality-specific temporal features and enables efficient inter-channel aggregation via lightweight graph attention. An anchor-free cascade predictor then refines action boundaries and class predictions in a two-stage design without dense proposals. Experiments on two real-world datasets demonstrate that SlimSTAD outperforms strong video-derived and sensory baselines by an average of 2.1 mAP, while significantly reducing GFLOPs, parameters, and latency, validating its effectiveness for real-world, edge-aware STAD deployment.

Source-free domain adaptive semantic segmentation aims at adapting a model trained on the source domain to the target domain without requiring access to the source data. Self-training has emerged as a leading approach to address this challenging problem. However, without robust denoising mechanisms to reduce the noise in pseudo labels, it still easily fall into biased estimates. Most existing methods address this issue by introducing novel architectures, but often at the cost of increased model complexity or reliance on additional input modalities. Different from previous studies, this article introduces UniSFDA , a unified multi-stage self-training framework that integrates cross-model transfer learning, uncertainty-aware pseudo label fusion, and intra-domain style augmentation, thereby enhancing both segmentation accuracy and test-time efficiency. Our proposed framework offers exceptional flexibility, with each component being independent and ready to be integrated into any existing self-training framework. Additionally, we investigate the performance of various representative segmentation models, including DeepLabv2, SegFormer, DFormer, and ViT-Adapter, within our framework. It is worth noting that UniSFDA is model-agnostic, allowing both source and target networks to be instantiated with arbitrary segmentation architectures, and thus readily benefiting from future advances in segmentation models. Experiments on the GTA5 \(\rightarrow\) Cityscapes and SYNTHIA \(\rightarrow\) Cityscapes benchmarks demonstrate the effectiveness of our framework. With DeepLabv2 (SegFormer) as the source model, UniSFDA establishes new state-of-the-art performance, achieving mIoU scores of 61.8% (65.4%) and 57.9% (59.6%) on the two benchmarks, respectively.

Precipitation nowcasting predicts future radar sequences based on current observations, which is a highly challenging task driven by the inherent complexity of the Earth system. Accurate nowcasting is of utmost importance for addressing various societal needs, including disaster management, agriculture, transportation, and energy optimization. As a complementary to existing non-autoregressive nowcasting approaches, we investigate the impact of prediction horizons on nowcasting models and propose SimCast, a novel training pipeline featuring a short-to-long term knowledge distillation technique coupled with a weighted MSE loss to prioritize heavy rainfall regions. Improved nowcasting predictions can be obtained without introducing additional overhead during inference. As SimCast generates deterministic predictions, we further integrate it into a diffusion-based framework named CasCast, leveraging the strengths from probabilistic models to overcome limitations such as blurriness and distribution shift in deterministic outputs. Extensive experimental results on three benchmark datasets validate the effectiveness of the proposed framework, achieving mean CSI scores of 0.452 on SEVIR, 0.474 on HKO-7, and 0.361 on MeteoNet, which outperforms existing approaches by a significant margin.

Traffic prediction is a crucial component of data management systems, leveraging historical data to learn spatio-temporal dynamics for forecasting future traffic and enabling efficient decision-making and resource allocation. Despite efforts to develop increasingly complex architectures, existing traffic prediction models often struggle to generalize across diverse datasets and contexts, limiting their adaptability in real-world applications. In contrast to existing traffic prediction models, large language models (LLMs) progress mainly through parameter expansion and extensive pre-training while maintaining their fundamental structures. In this paper, we propose ST-LLM+, the graph enhanced spatio-temporal large language models for traffic prediction. Through incorporating a proximity-based adjacency matrix derived from the traffic network into the calibrated LLMs, ST-LLM+ captures complex spatio-temporal dependencies within the traffic network. The Partially Frozen Graph Attention (PFGA) module is designed to retain global dependencies learned during LLMs pre-training while modeling localized dependencies specific to the traffic domain. To reduce computational overhead, ST-LLM+ adopts the LoRA-augmented training strategy, allowing attention layers to be fine-tuned with fewer learnable parameters. Comprehensive experiments on real-world traffic datasets demonstrate that ST-LLM+ outperforms state-of-the-art models. In particular, ST-LLM+ also exhibits robust performance in both few-shot and zero-shot prediction scenarios. Additionally, our case study demonstrates that ST-LLM+ captures global and localized dependencies between stations, verifying its effectiveness for traffic prediction tasks.

Spatio-temporal graph neural networks have proven efficacy in capturing complex dependencies for urban computing tasks such as forecasting and kriging. Yet, their performance is constrained by the reliance on extensive data for training on a specific task, thereby limiting their adaptability to new urban domains with varied task demands. Although transfer learning has been proposed to remedy this problem by leveraging knowledge across domains, the cross-task generalization still remains under-explored in spatio-temporal graph transfer learning due to the lack of a unified framework. To bridge the gap, we propose Spatio-Temporal Graph Prompting (STGP), a prompt-based framework capable of adapting to multi-diverse tasks in a data-scarce domain. Specifically, we first unify different tasks into a single template and introduce a task-agnostic network architecture that aligns with this template. This approach enables capturing dependencies shared across tasks. Furthermore, we employ learnable prompts to achieve domain and task transfer in a two-stage prompting pipeline, facilitating the prompts to effectively capture domain knowledge and task-specific properties. Our extensive experiments demonstrate that STGP outperforms state-of-the-art baselines in three tasks-forecasting, kriging, and extrapolation-achieving an improvement of up to 10.7%.

Air quality prediction and modelling plays a pivotal role in public health and environment management, for individuals and authorities to make informed decisions. Although traditional data-driven models have shown promise in this domain, their long-term prediction accuracy can be limited, especially in scenarios with sparse or incomplete data and they often rely on black-box deep learning structures that lack solid physical foundation leading to reduced transparency and interpretability in predictions. To address these limitations, this paper presents a novel approach named Physics guided Neural Network for Air Quality Prediction (AirPhyNet). Specifically, we leverage two well-established physics principles of air particle movement (diffusion and advection) by representing them as differential equation networks. Then, we utilize a graph structure to integrate physics knowledge into a neural network architecture and exploit latent representations to capture spatio-temporal relationships within the air quality data. Experiments on two real-world benchmark datasets demonstrate that AirPhyNet outperforms state-of-the-art models for different testing scenarios including different lead time (24h, 48h, 72h), sparse data and sudden change prediction, achieving reduction in prediction errors up to 10\%. Moreover, a case study further validates that our model captures underlying physical processes of particle movement and generates accurate predictions with real physical meaning. The code is available at: https://github.com/kethmih/AirPhyNet

Source-free domain adaptive semantic segmentation has gained increasing attention recently. It eases the requirement of full access to the source domain by transferring knowledge only from a well-trained source model. However, reducing the uncertainty of the target pseudo labels becomes inevitably more challenging without the supervision of the labeled source data. In this work, we propose a novel asymmetric two-stream architecture that learns more robustly from noisy pseudo labels. Our approach simultaneously conducts dual-head pseudo label denoising and cross-modal consistency regularization. Towards the former, we introduce a multimodal auxiliary network during training (and discard it during inference), which effectively enhances the pseudo labels' correctness by leveraging the guidance from the depth information. Towards the latter, we enforce a new cross-modal pixel-wise consistency between the predictions of the two streams, encouraging our model to behave smoothly for both modality variance and image perturbations. It serves as an effective regularization to further reduce the impact of the inaccurate pseudo labels in source-free unsupervised domain adaptation. Experiments on GTA5 → Cityscapes and SYNTHIA → Cityscapes benchmarks demonstrate the superiority of our proposed method, obtaining the new state-of-the-art mIoU of 57.7% and 57.5%, respectively.

Tube internal erosion, which corresponds to its wall thinning process, is one of the major safety concerns for tubes. Many sensing technologies have been developed to detect a tube wall thinning process. Among them, fiber Bragg grating (FBG) sensors are the most popular ones due to their precise measurement properties. Most of the current works focus on how to design different types of FBG sensors according to certain physical laws and only test their sensors in controlled laboratory conditions. However, in practice, an industrial system usually suffers from harsh and dynamic environmental conditions, and FBG signals are affected by many unpredictable factors. Consequently, the FBG signals have more fluctuations and are polluted by noises. Hence, the signals no longer directly follow the assumed physical laws and their proposed thinning detection mechanisms no longer work. Targeting at this, this article develops a data-driven model for FBG signal feature extraction and tube wall thickness monitoring using data analytic techniques. In particular, we develop a spatiotemporal model to describe dynamic FBG signals and extract features related to thickness. By taking physical law as guideline, we trace the relationship between the extracted features and the tube wall thickness, based on which we construct an online statistical monitoring scheme for tube wall thinning process. We use both laboratory test and field trial experiment to demonstrate the efficacy and efficiency of the proposed scheme.Note to Practitioners—This article is motivated by the real industrial needs of inner erosion detection of tubes in harsh environment. Most of the current research works focus on designing various sensing apparatuses based on fiber Bragg grating (FBG) sensors for nondestructive erosion detection. These apparatuses prove to be able to collect signals reflecting tube wall thickness in static and controllable laboratory environment qualitatively. However, in reality, the industrial environment, which is impacted by many changing factors, is dynamic and uncontrollable. Consequently, the signals collected by these FBG apparatuses would have larger variations that mask the signals related to thickness. Furthermore, current methods have neither mentioned how to process their collected data to capture the unnoticeably slow but accumulative erosion information efficiently nor constructed online monitoring algorithms to detect the tube wall thinning process based on the collected signals quantitatively. Built upon their apparatuses but targeting at their unsolved challenges, we propose a novel data-driven approach for FBG signal analysis that can remove the environmental influence and extract features only related to tube wall thickness, and using the extracted features, we construct a statistical process control scheme to monitor tube wall thickness and detect erosion in real time efficiently.

Previous work on the spatio-temporal index often adopts a simple linear model to predict the future positions of moving objects, which may generate numerous errors for complex road networks and fast moving objects. In this paper, we propose NEIST, a neural-enhanced index to process spatio-temporal queries with enhanced efficiency and accuracy, by intelligently leveraging the movement patterns among moving objects. NEIST applies a Recurrent Neural Network (RNN) model to predict future positions of moving objects based on observed trajectories. To reduce the prediction overhead, a suffix-tree is further built to index trajectories with similar suffixes, and thus similar objects within a given similarity bound are grouped together to share the same prediction result. A prediction result in NEIST represents possible positions of a group of moving objects in the next t time slots. Inside each time slot, traditional linear prediction model is then adopted and a TPR-Tree is built to support spatio-temporal queries. We use Singapore and Porto taxi trajectory datasets to evaluate NEIST. Compared to previous approaches, NEIST achieves a much more efficient query performance and is able to produce about 70 percent more accurate results.

While analysis of urban commuting data has a long and demonstrated history of providing useful insights into human mobility behavior, such analysis has been performed largely in offline fashion and to aid medium-to-long term urban planning. In this work, we demonstrate the power of applying predictive analytics on real-time mobility data, specifically the smart-card generated trip data of millions of public bus commuters in Singapore, to create two novel and "live" smart city services. The key analytical novelty in our work lies in combining two aspects of urban mobility: (a) conformity: which reflects the predictability in the aggregated flow of commuters along bus routes, and (b) regularity: which captures the repeated trip patterns of each individual commuter. We demonstrate that the fusion of these two measures of behavior can be performed at city-scale using our BuScope platform, and can be used to create two innovative smart city applications. The Last-Mile Demand Generator provides O(mins) lookahead into the number of disembarking passengers at neighborhood bus stops; it achieves over 85% accuracy in predicting such disembarkations by an ingenious combination of individual-level regularity with aggregate-level conformity. By moving driverless vehicles proactively to match this predicted demand, we can reduce wait times for disembarking passengers by over 75%. Independently, the Neighborhood Event Detector uses outlier measures of currently operating buses to detect and spatiotemporally localize dynamic urban events, as much as 1.5 hours in advance, with a localization error of ~450 meters.

Malfunctions in machines require equipment engineers to conduct fault diagnostic. The fault diagnostics is traditionally reliant on the skills and experiences of the equipment operators and maintenance engineers heavily, which creates an unnecessary technical barrier and results in extra cost in downtime cost and operation overhead. Meanwhile, there have been rich machine logs (sensory readings, performance logs, system logs, context data, process data) of the machine as well as the maintenance data and post-service reports. Such data provide an opportunity of leveraging intelligent data-driven technologies to reduce the maintenance cost through developing automated solutions on machine fault diagnostics and prognostic for critical component failures.In this paper, a data driven framework is proposed for machine diagnostics and prognostics to relieve the maintenance cost and increase the efficiency. It has been validated with real-world big data from complex vending machines. The proposed framework addresses the data size issue effectively by deriving applicable features and subsequently sustaining the top attributable ones only in the model. An accurate data labeling methodology is developed for supervised learning via comparing the serial number of target components in the adjacent dates. Two predictive models have been developed in this work whereby the first one is in the domain of binary classification for diagnostics, and the second one is a generalized two-stage prognostics model for multi-class classification for preventive maintenance. Cross-validated simulation results have shown that our developed diagnostics model can achieve above 80% accuracy in terms of precision, recall, and F-measure. It has also been shown that the proposed two-stage prognostics framework can outperform the conventional one-stage multiclass prediction models.

Trip time prediction is an important problem. Taxi passengers often want to know when they will arrive at their destinations. We design a method of predicting taxi trip time by finding historical similar trips. Trips are clustered based on origin, destination, and start time. Then similar trips are mapped to road networks to find frequent sub-trajectories that are used to model travel time of the various parts of the routes. Experimental results show this method is effective.

Taxi waiting queues or passenger waiting queues usually reflect the imbalance between taxi supply and demand, which consequently decrease a city’s trac system productivity and commuters’ satisfaction. In this paper, we present a queue detection and analysis system to conduct analytics on both taxi and passenger queues. The system utilizes the event-driven taxi traces and the taxi state transition knowledge to detect queue locations at a coordinate level and subsequently identify 4 di↵erent types of queue context (e.g., only passengers queuing or only taxis queuing). More specifically, it adopts the novel and easy-to-implement algorithms to selectively extract taxi pickup events and their critical features. The extracted taxi pickup locations are then used to detect queue locations, and the extracted critical features are used to infer queue context. The extensive empirical evaluations, which run on daily 12.4 million taxi trace records from nearly 15000 taxis in Singapore, demonstrate the high accuracy and stability of the queue analytics results. Finally, we discuss the real world deployment issues and the gained insights from the queue analysis results.

Personal mobile data are being extensively collected by various service providers, in the form of data stream. Most service providers promise their customers for not misusing their data by paper-based agreement. However, the customers have no way to know whether the agreements are strictly followed or not, unless any scandals of private data misuse are revealed. To guarantee the correct use of customers' personal data and assure them of the service safety, system-level data privacy control between the data owners (i.e., Customers) and the data users (i.e., Service providers) is in compelling need. Inspired by the concept of Hippocratic data management, we design and implement a system, Hip Stream to systemically enforce different Hippocratic principles to preserve data providers' privacy when they send their data stream for services. In this paper, we describe the architecture of the Hip Stream system and demonstrate how it meets those privacy principles.

The Electronic Road Pricing (ERP) system was implemented by the Land Transport Authority of Singapore to control traffic by road pricing since 1998. To better understand the traffic condition and improve the pricing scheme, the government initiated the next generation ERP (ERP 2) project, which aims to use the Global Navigation Satellite System (GNSS) collecting positional data from vehicles for analysis. However, most drivers fear of being monitored once the government installs the devices in their vehicles to collect GPS data. The existing data stream management systems (DSMS) centralize both data management and privacy control at server site. This framework assumes DSMS server is secure and trustable, and protects providers' data from illegal access by data users. In ERP 2, the DSMS server is maintained by the government, i.e., data user. Thus, the existing framework is not adoptable. We propose a novel framework in which privacy protection is pushed to data provider site. By doing this, the system could be safer and more efficient. Our framework can be used for the situations such as ERP 2, i.e., data providers would like to control their own privacy policies and/or the workload of DSMS server needs to be reduced.

Big data management and analysis has become a hot topic in academic and industrial research. In fact, a large portion of big data in service today are initially streaming data. To preserve the privacy of such data that are collected from data streams, the most efficient way is to control the process of data collection according to corresponding privacy polices. In this paper, we design a framework to support data stream management with privacy-preserving capabilities. In particular, we focus on two premier principles of data privacy, limited disclosure and limited collection. With these two principles guaranteed, the archived data will not necessarily be checked for privacy protection, before analysis and other operations can be done.

In mobile sensor networks (MSNs), it is common for the base station to issue {\em data acquisition} queries requesting for data to be sensed from specific regions of the data space. Such kind of queries are especially important in MSNs for reconnaissance and disaster rescue applications. In this paper, we investigate how multiple data acquisition queries can be answered quickly in sparse mobile sensor networks. Because of the sparseness and mobility, the number of sensors is limited, the connection is intermittent and the topology is unpredictable. To effectively handle the above challenges, we design distributed schemes where mobile sensors strategically relocate themselves to proper locations to collaboratively facilitate efficient query processing and enable sharing over space and time. We first propose a novel scheme, {\em Dynamic}, that enables queries to share resources at runtime while sensors are greedily relocated to benefit the processing of each query. We also design another scheme, {\em aMST}, that optimizes a batch of queries as a whole and utilizes a Minimum Steiner Tree to guide the execution of all queries in the batch. In addition, a parameter is defined to guide the selection of the most appropriate scheme to adapt to the environment. Our extensive performance study shows the effectiveness of our proposed schemes.

When there are multiple queries posed to the resource-constrained wireless sensor network, it is critical to process them efficiently. In this paper, we propose a two-tier multiple query optimization (TTMQO) scheme. The first tier, called base station optimization, adopts a cost-based approach to rewrite a set of queries into an optimized set that shares the commonality and eliminates the redundancy among the queries in the original set. The optimized queries are then injected into the wireless sensor network. In the second tier, called in-network optimization, our scheme efficiently delivers query results by taking advantage of the broadcast nature of the radio channel and sharing the sensor readings among similar queries over time and space at a finer granularity. Our experimental results indicate that our proposed TTMQO scheme offers significant improvements over the traditional single query optimization technique.

Our goal is to design a light-weight but effective scheme to support multiple data acquisition and aggregation queries in a wireless sensor network, in order to minimize the number of radio transmissions. Apart from being much more powerful than sensor nodes, the base station is also the interface of a wireless sensor network. Thus, we use the base station as a filter to reduce duplicate data accesses from the sensor network, and as a screen to hide the query dynamics as much as possible. We design a two-tier optimization scheme, base station optimization and in-network optimization.