Accurate traffic forecasting is critical in improving safety, stability, and efficiency of intelligent transportation systems. Despite years studies, accurate prediction still faces the following challenges, including modeling dynamics data along both temporal spatial dimensions, capturing periodicity heterogeneity data, problem more difficult for long-term forecast. In this paper, we propose an Attention based Spatial-Temporal Graph Neural Network (ASTGNN) forecasting. Specifically,...

We present the design, implementation, and evaluation of AirCloud -- a novel client-cloud system for pervasive personal air-quality monitoring at low cost. At frontend, we create two types Internet-connected particulate matter (PM2:5) monitors AQM miniAQM, with carefully designed mechanical structures optimal air-flow. On cloud-side, an analytics engine that learn models based on fusion sensor data. This is used to calibrate AQMs mini-AQMs in real-time, infer PM2:5 concentrations. evaluate...

Trajectory similarity computation is fundamental functionality with many applications such as animal migration pattern studies and vehicle trajectory mining to identify popular routes similar drivers. While a continuous curve in some spatial domain, e.g., 2D Euclidean space, trajectories are often represented by point sequences. Existing approaches that compute based on matching suffer from the problem they treat two different sequences differently even when represent same trajectory. This...

As a fundamental problem in social influence propagation analysis, learning parameters has been extensively investigated. Most of the existing methods are proposed to estimate probability for each edge networks. However, they cannot effectively learn all edges due data sparsity, especially without sufficient observed propagation. Different from conventional methods, we introduce novel embedding problem, which is nodes rather than edges. Nodes represented as vectors low-dimensional space, and...

Travel time estimation of a given route with respect to real-time traffic condition is extremely useful for many applications like planning. We argue that it even more estimate the travel distribution, from which we can derive expected as well uncertainty. In this paper, develop deep generative model - DeepGTT learn distribution any by conditioning on traffic. interprets generation using three-layer hierarchical probabilistic model. first layer, present two techniques, amortization and...

In this work, we propose a robust road network representation learning framework called Toast, which comes to be cornerstone boost the performance of numerous demanding transport planning tasks. Specifically, first traffic context aware skip-gram module incorporate auxiliary tasks predicting target segment. Furthermore, trajectory-enhanced Transformer that utilizes trajectory data extract traveling semantics on networks. Apart from obtaining effective segment representations, also enables us...

In this paper, we study the problem of predicting most likely traveling route on road network between two given locations by considering real-time traffic. We present a deep probabilistic model-DeepST-which unifies three key explanatory factors, past traveled route, impact destination and traffic for decision. DeepST explains generation next conditioning representations factors. To enable effectively sharing statistical strength, propose to learn K-destination proxies with an adjoint...

Air quality is one of the most important environmental concerns in world, and it has deteriorated substantially over past years many countries. For example, Chinese Academy Social Sciences reports that problem haze fog China hitting a record level, currently suffering from worst air pollution. Among various causal factors quality, particulate matter with diameter 2.5 micrometers or less (i.e., PM2.5) very factor; governments people are increasingly concerned concentration PM2.5. In cities,...

Urban grid prediction can be applied to many classic spatial-temporal tasks such as air quality prediction, crowd density and traffic flow which is of great importance smart city building. In light its practical values, methods have been developed for it achieved promising results. Despite their successes, two main challenges remain open: (a) how well capture the global dependencies (b) effectively model multi-scale correlations? To address these challenges, we propose a novel method—...

As a fundamental component in location-based services, inferring the relationship between points-of-interests (POIs) is very critical for service providers to offer good user experience business owners and customers. Most of existing methods inference are not targeted at POI, thus failing capture unique spatial characteristics that have huge effects on POI relationships. In this work we propose PRIM tackle multiple relation types. features four novel components, including weighted relational...

Abstract Collecting energy from body movements to supply power for equipment is an important method rapidly develop wearable intelligent equipment. This places high requirements, such as being lightweight and stable, on a nanogenerator. Herein, bioinspired helical triboelectric nanogenerator (BH‐TENG) that realizes acquisition through the intensity frequency of movement designed based structure DNA. The facilitates more contact areas provides increased stability complete generation process...

By optimizing the aerodynamic shape parameters, performance of vehicle becomes better as drag decreases. The driving stability also lift This research presents optimization which employs multi-variable parametric model and iterative optimal approach to reduce lift. For studies with computational fluid dynamics simulations, a surface grid was used morph enhance mesh quality by linear deformation exterior surfaces. method employed radial basis function model, integrated multi-software provides...

Similar trajectory search is a crucial task that facilitates many downstream spatial data analytic applications. Despite its importance, of the current literature focus solely on trajectory’s similarity while neglecting temporal information. Additionally, few papers use both and features based their approach traditional point-to-point comparison. These methods model importance aspect with only single, pre-defined balancing factor for all trajectories, even though relative balance can change...

Over the past decade, smartphones have become indispensable personal mobile devices, experiencing a remarkable surge in software apps. These apps empower users to seamlessly connect with various internet services, such as social communication and online shopping. Accurately predicting smartphone app usage can effectively improve user experience optimize resource utilization. However, existing models often treat prediction classification problem, which suffers from issues of imbalance...

This paper presents the design and implementation of a novel user-initiated indoor localization system called QiLoc. QiLoc is simple yet effective way to accurately locate identify occupants Qi-compatible devices inside buildings. The composed Stations Server. A Station usually embedded or under desk/table. By using Qi wireless charging protocol, it can extract unique ID device, thus locating occupant. Server maintains location information all occupants, provides set APIs via standard web...

Dispersion of vehicle exhaust gas is a primary source air pollution in urban areas. Thus, it has become an important subject the automotive field. This paper consists two parts. First, fastback MIRA model was selected as study object and standard<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" id="M1"><mml:mi>κ</mml:mi><mml:mtext>-</mml:mtext><mml:mi>ε</mml:mi></mml:math>two-equation turbulence used. The simulation results were compared analyzed with experimental data. feasibility...

Recommender systems play a vital role in modern web services. In typical recommender system, we are given set of observed user-item interaction records and seek to uncover the hidden behavioral patterns users from these historical interactions. By exploiting patterns, aim discover users' personalized tastes recommend them new items. Among various types recommendation methods, latent factor collaborative filtering models have dominated field. this paper, develop unified view for existing...

In this study, we introduce a novel framework called Toast for learning general-purpose representations of road networks, along with its advanced counterpart DyToast, designed to enhance the integration temporal dynamics boost performance various time-sensitive downstream tasks. Specifically, propose encode two pivotal semantic characteristics intrinsic networks: traffic patterns and traveling semantics. To achieve this, refine skip-gram module by incorporating auxiliary objectives aimed at...

With human action anticipation becoming an essential tool for many practical applications, there has been increasing trend in developing more accurate models recent years. Most of the existing methods target standard datasets, which they could produce promising results by learning action-level contextual patterns. However, over-simplified scenarios datasets often do not hold reality, hinders them from being applied to real-world applications. To address this, we propose a scene-graph-based...

Time series forecasting is essential for our daily activities and precise modeling of the complex correlations shared patterns among multiple time improving performance. Spatial-Temporal Graph Neural Networks (STGNNs) are widely used in multivariate tasks have achieved promising performance on real-world datasets their ability to model underlying spatial temporal dependencies. However, existing studies mainly focused comprising only a few hundred sensors due heavy computational cost memory...