ABSTRACT The rapid advancement of mobile edge computing (MEC) has led to a growing need for executing complex in‐vehicle applications at the endpoint. These applications are often modeled as workflows with clearly defined task dependencies. In MEC environments, computation offloading plays a crucial role in enhancing Quality of Service (QoS). Nevertheless, most existing offloading strategies primarily address independent tasks, paying little attention to the intricate dependencies among tasks. Moreover, key factors such as budgets and privacy protection are not well tackled. Hence, in this paper, we propose a computation offloading model for workflow applications in mobile edge computing, where the makespan is optimized under budget and privacy constraints. To address it, we introduce an intelligent scheduling method, which consists of task offloading and resource preparation stages, and offloads tasks from the end to the edge and clouds with awareness of task urgency, device computing capabilities, economic costs, and privacy protection. Simulation experiments are conducted based on realistic datasets. Results demonstrate that the proposed method can improve task execution efficiency and ensure data security, thereby enhancing the performance of workflow applications in mobile edge computing scenarios.

Event-based semantic segmentation has great potential in autonomous driving and robotics due to the advantages of event cameras, such as high dynamic range, low latency, and low power cost. Unfortunately, current artificial neural network (ANN)-based segmentation methods suffer from high computational demands, the requirements for image frames, and massive energy consumption, limiting their efficiency and application on resource-constrained edge/mobile platforms. To address these problems, we introduce SLTNet, a Spike-driven Lightweight Transformer-based Network designed for event-based semantic segmentation. Specifically, SLTNet is built on efficient spike-driven convolution blocks (SCBs) to extract rich semantic features while reducing the model’s parameters. Then, to enhance the long-range contextual feature interaction, we propose novel spike-driven transformer blocks (STBs) with binary mask operations. Based on these basic blocks, SLTNet employs a high-efficiency single-branch architecture while maintaining the low energy consumption of the Spiking Neural Network (SNN). Finally, extensive experiments on DDD17 and DSEC-Semantic datasets demonstrate that SLTNet outperforms state-of-the-art (SOTA) SNN-based methods by at most 9.06% and 9.39% mIoU, respectively, with extremely 4.58× lower energy consumption and 114 FPS inference speed. Our code is open-sourced and available at https://github.com/longxianlei/SLTNet-v1.0.

Representation learning of road networks is essential for various downstream traffic-related tasks, as road network contain multi-modal data with rich information, and the learned embeddings can be directly used in machine learning models. However, due to the dynamic changes in road networks with respect to topology and associated data, as well as the local and long-range dependency caused by complex mobility semantics, learning robust and effective representations remains challenging. To this end, we exploit the properties of the road network and the mobility semantics embedded in trajectories, and propose a novel S emantic- E nhanced G raph C ontrastive L earning (SE-GCL) framework, for learning general-purpose embeddings of road networks. Specifically, in this framework, we propose (1) a multi-modal feature embedding module to capture both the attribute and visual information of road segments, (2) a semantic-enhanced graph augmentation strategy to simulate topological changes and data missing in the road network, and (3) a semantic-enhanced contrastive optimization module that leverages geo-locality and mobility semantics to guide representation learning. Extensive experiments are conducted on two real-world road networks with three representative downstream tasks. The result demonstrate that SE-GCL yields more robust and effective representations, outperforming the state-of-the-art baselines. The source code is available at https://github.com/csjiezhao/SE-GCL .

Region representations (also called embeddings) are useful for various urban computing tasks. While graph-based region representation learning methods have shown outstanding performance, they encounter two major challenges: 1) the pervasive data noise and missing data can affect the quality of the constructed region graphs; and 2) high-order relationships (i.e., group-wise relationships) among regions are often insufficiently modeled and sometimes entirely overlooked. To this end, we proposeHyperRegion, an unsupervised region representation learning framework that integrates graph and hypergraph contrastive learning to learn comprehensive region embeddings from multi-modal data. Built upon a region hybrid graph network, this framework models both pair-wise and group-wise dependencies involving POI semantics, mobility patterns, geographic neighbors, and visual semantics. To mitigate the impact of data noise and missing data, graph and hypergraph contrastive learning are performed in parallel, and a cross-module contrast is further introduced to facilitate information exchange and collaboration. Extensive experiments on real-world datasets across three downstream tasks demonstrate thatHyperRegionoutperforms all baselines, particularly improving check-in prediction by reducing MAE and RMSE by approximately 8.5% and 8.2%, respectively, and increasing$R^{2}$by about 7%.

While a variety of innovative travel modes, such as taxi service and ridesourcing service, have been launched to improve the transportation efficiency, people still encounter travel problems in real life. The major cause is the imbalance between transportation supply and demand. To strike a balance, it is well-recognized that an accurate and timely passenger demand prediction model is the foundation to enable high-level human intelligence (i.e., taxi drivers) or machine intelligence (i.e., ride-hailing platforms) to allocate resources in advance. Although quite a lot of deep models have been designed to model the complicated spatial and temporal dependencies in a data-driven way, they focus on the demand prediction of a single mode and ignore the fact that passengers may shift between different modes, especially between taxis and ridesourcing cars. In this paper, we target a co-prediction problem that considers the prediction of taxi and ridesourcing as two coupled and associated tasks, and propose a novel Temporal and Spatial Intertwined Network (TSIN) that consists of two twin components and an intertwined component. Each twin in the TSIN model is able to extract spatial and temporal dependencies from its corresponding travel mode separately (i.e., intra-mode features), and the in-between intertwined component is designed to bridge the twins and allow them to exchange information (i.e., inter-mode features), thus enabling better prediction. We first evaluate our model on four real-world datasets. Results demonstrate the outstanding performance of our model and the necessity to take into account the influence between modes. Based on an additional demand data from bike in NYC, we then discuss the generalizability in coupling more transportation modes. Further results demonstrate that our proposed intertwined neural network is highly flexible and extendable, and can yield better prediction performance.

Trip purpose is a meaningful aspect of travel behaviour for the understanding of urban mobility. However, it is non-trivial to automatically obtain trip purposes. On one hand, trip purposes are naturally diverse and complicated, but the available predictive data sources are limited in real-world scenarios. On the other hand, since trip purpose labeling is costly and the development levels of cities are unbalanced, it is infeasible to access large-scale labeled data in less developed cities to train advanced prediction models. To narrow the gaps, this article presents A new Graph Embedding Network and active Domain Adaptation based framework (AGENDA) that only requires open data sources and is capable of predicting in both label-rich cities and label-scarce cities. Specifically, in label-rich source cities, we first use the vehicle’s GPS trajectory and open POI check-ins to augment trip contexts. Then we establish a supervised graph embedding network with two attention mechanisms to extract the passenger’s latent activity semantics and a classifier to predict trip purpose. To enable the prediction in label-scarce target cities, we further devise an active domain adaptation framework, in which adversarial domain adaptation is used to transfer the source-learned knowledge, and active learning is used to integrate human intelligence in the model training. A group of experiments are conducted with real-world datasets in Beijing and Shanghai. Evaluation results demonstrate that the proposed framework significantly outperforms existing trip purpose prediction algorithms, and could make accurate trip purpose prediction in label-scarce cities with much fewer labeling efforts.

The proliferation of multimodal big data in cities provides unprecedented opportunities for modeling and forecasting urban problems, such as crime prediction and house price prediction, through data-driven approaches. A fundamental and critical issue in modeling and forecasting urban problems lies in identifying suitable spatial analysis units, also known as city region partition. Existing works rely on subjective domain knowledge for static partitions, which is general and universal for all tasks. In fact, different tasks may need different city region partitions. To address this issue, we propose JLPR , a task-oriented framework for J oint L earning of region P artition and R epresentation. To make partitions fit tasks, JLPR integrates the region partition into the representation model training and learns region partitions using the supervision signal from the downstream task. We evaluate the framework on two prediction tasks (i.e., crime prediction and housing price prediction) in Chicago. Experiments show that JLPR consistently outperforms state-of-the-art partitioning methods in both tasks, which achieves above 25% and 70% performance improvements in terms of mean absolute error for crime prediction and house price prediction tasks, respectively. Additionally, we meticulously undertake three visualization case studies, which yield profound and illuminating findings from diverse perspectives, demonstrating the remarkable effectiveness and superiority of our approach.

Abstract Virtual reality technology is also known as VR technology. Accurately acquiring the three‐dimensional model of the point cloud on the surface of an object is a key part of virtual reality technology, and is also a research hotspot of virtual reality technology. The purpose of this article is to study the object surface point cloud computing algorithm based on virtual reality technology and its application problems and to solve the problems in the virtual reality technology object surface point cloud data algorithm, by reconstructing the geometry of the object surface in VR technology. The calculation rules at the time of data acquisition have been carried out in detailed research on the point cloud space carving algorithm based on the surface of multi‐eye visual objects. The research results show that the point cloud space carving algorithm on the surface of the virtual reality technology has a high degree of independence and the algorithm is very flexible. Compared with other algorithms, it reduces the operation steps by 35%, the 3D effect is increased by 21%, and the operation cost is reduced by 30%. The object surface point cloud computing algorithm based on virtual reality technology can greatly improve the computing efficiency, compared with the traditional two‐dimensional visual image integration algorithm, the efficiency is increased by 35%, and the computing accuracy in virtual reality technology is the same as the traditional two The accuracy of the integrated algorithm of dimensional visual pictures is improved by 44% compared to the accuracy. The experimental results verify the feasibility of applying the spatial carving algorithm to the three‐dimensional reconstruction of point cloud data on the surface of virtual reality technology objects.

In personal context recognition many solutions rely on supervised learning that uses sensor data collected from the users' mobile devices. However, the recognition performance is significantly affected by the annotations’ quality. The problem lies in the fact that the annotator in such scenarios is usually the user herself which is not an expert and thus provides a significant amount of incorrect labels, while existing solutions can only tolerate a small fraction of mislabels. Our solution is Skeptical Learning, a framework for interactive machine learning where the machine uses all its available knowledge to check the correctness of its own and the user labeling. This allows to have a uniform confidence measure to be used when a contradiction arises that applies to both the annotator and the machine. The criteria of success is an improvement of the final recognition accuracy with respect to traditional supervised approaches.