Forecasting traffic flow is an important task in urban areas, and a large number of methods have been proposed for prediction. However, most the existing follow general technical route to aggregate historical information spatially temporally. In this paper, we propose different approach Our major motivation more effectively incorporate various intrinsic patterns real-world flows, such as fixed spatial distributions, topological correlations, temporal periodicity. Along line, novel...

Graph Neural Networks (GNNs) have been widely applied in the semi-supervised node classification task, where a key point lies how to sufficiently leverage limited but valuable label information. Most of classical GNNs solely use known labels for computing loss at output. In recent years, several methods designed additionally utilize input. One part augment features via concatenating or adding them with one-hot encodings labels, while other optimize graph structure by assuming neighboring...

Hard landing is a severe accident during the flight phase, which threats aircraft architecture and passengers' safety. This study proposed model named LSTM for hard prediction, provides advanced warning to take proper measures. The unique structure of makes it have superior capability capture long temporal dependency time series QAR data forecasting. Experiments were conducted using A320 dataset consisting 853 flights 1082 normal flights. Comparing performance other tradition prediction...

Information propagation on social networks could be modeled as cascades, and many efforts have been made to predict the future popularity of cascades. However, most existing research treats a cascade an individual sequence. Actually, cascades might correlated with each other due shared users or similar topics. Moreover, preferences semantics are usually continuously evolving over time. In this paper, we propose continuous-time graph learning method for prediction, which first connects...

Given a sequence of sets, where each set contains an arbitrary number elements, temporal sets prediction aims to predict which elements will appear in the subsequent set. Existing methods for are developed on sophisticated components (e.g., recurrent neural networks, attention or gating mechanisms, and graph networks), inevitably increase model complexity due more trainable parameters higher computational costs. Moreover, involved nonlinear activation may contribute little even degrade...

Travel time estimation (TTE) is a fundamental and challenging problem for navigation travel planning. Though many efforts have been devoted to this task, most of the previous research has focused on extracting useful features routes improve accuracy. In our opinion, key issue TTE how handle rich spatiotemporal information underlying route model multi-faceted factors that affect time. Along line, we propose representation learning framework divides into three sequences: trajectory sequence...

Highway traffic volume prediction is critical for urban planning and relieving stress. In recent years, many researches have been focusing on it by using machine learning algorithms. However, different methodologies are required evaluation metrics the performance of single model limited. this paper, we present our models which respectively inspired from KNN, SVR, MLP RNN, to predict highway volume, aiming at metrics. Based revised models, stacking used can boost predictive accuracy combining...

A hard landing occurs when an aircraft or spacecraft hits the ground with a greater vertical speed and force than in normal landing, which can cause serious damage to even put passengers great danger. Predicting ahead is valuable, however, it has not been fully investigated. Existing studies mainly apply statistical machine learning algorithms feature engineering on small-scale data sets. Such methods are primarily dependent constructing features by experts. In addition, since samples...

Origin-Destination demand prediction is a fundamental and important task in the urban transportation system. It more challenging complex than region since it needs to predict traffic for each pair of regions rather single region, which means <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$N^{2}$ </tex-math></inline-formula> time series need be predicted given notation="LaTeX">$N$ stations. Most existing...

Dynamic graph learning has attracted much attention in recent years due to the fact that most of real-world graphs are dynamic and evolutionary. As a result, many methods have been proposed cope with changes node states over time. Among these studies, critical issue is how update representations nodes when new temporal events observed. In this paper, we provide novel memory structure - Memory Map (MemMap) for problem. MemMap an adaptive evolutionary latent space, where each cell corresponds...