Simultaneous Past and Current Social Interaction-aware Trajectory Prediction for Multiple Intelligent Agents in Dynamic Scenes

Trajectory prediction of multiple agents in a crowded scene is an essential component in many applications, including intelligent monitoring, autonomous robotics, and self-driving cars. Accurate agent trajectory prediction remains a significant challenge because of the complex dynamic interactions among the agents and between them and the surrounding scene. To address the challenge, we propose a decoupled attention-based spatial-temporal modeling strategy in the proposed trajectory prediction method. The past and current interactions among agents are dynamically and adaptively summarized by two separate attention-based networks and have proven powerful in improving the prediction accuracy. Moreover, it is optional in the proposed method to make use of the road map and the plan of the ego-agent for scene-compliant and accurate predictions. The road map feature is efficiently extracted by a convolutional neural network, and the features of the ego-agent’s plan is extracted by a gated recurrent network with an attention module based on the temporal characteristic. Experiments on benchmark trajectory prediction datasets demonstrate that the proposed method is effective when the ego-agent plan and the the surrounding scene information are provided and achieves state-of-the-art performance with only the observed trajectories.

Spatio-Temporal Modeling and Simulation of Asian Soybean Rust Based on Fuzzy System

Mato Grosso, Brazil, is the largest soy producer in the country. Asian Soy Rust is a disease that has already caused a lot of damage to Brazilian agribusiness. The plant matures prematurely, hindering the filling of the pod, drastically reducing productivity. It is caused by the Phakopsora pachyrhizi fungus. For a plant disease to establish itself, the presence of a pathogen, a susceptible plant, and favorable environmental conditions are necessary. This research developed a fuzzy system gathering these three variables as inputs, having as an output the vulnerability of the region to the disease. The presence of the pathogen was measured using a diffusion-advection equation appropriate to the problem. Some coefficients were based on the literature, others were measured by a fuzzy system and others were obtained by real data. From the mapping of producing properties, the locations where there are susceptible plants were established. And the favorable environmental conditions were also obtained from a fuzzy system, whose inputs were temperature and leaf wetness. Data provided by IBGE, INMET, and Antirust Consortium were used to fuel the model, and all treatments, tests, and simulations were carried out within the Matlab® environment. Although Asian Soybean Rust was the chosen disease here, the model was general in nature, so could be reproduced for any disease of plants with the same profile.

Spatio-Temporal Modeling of Ozone Distribution in Tehran, Iran Based on Neural Network and Geographical Information System

Background: Air pollution is one of the most important causes of respiratory diseases that people face in big cities today. Suspended particulates, carbon monoxide, sulfur dioxide, ozone, and nitrogen dioxide are the five major pollutants of air that pose many problems to human health. We aimed to provide an approach for modeling and analyzing the spatiotemporal model of ozone distribution based on Geographical Information System (GIS). Methods: In the first step, by considering the accuracy of different interpolation methods, the Inverse distance weighted (IDW) method was selected as the best interpolation method for mapping the concentration of ozone in Tehran, Iran. In the next step, according to the daily data of Ozone pollutants, the daily, monthly, and annual mean concentrations maps were prepared for the years 2015, 2016, and 2017. Results: Spatial and temporal analysis of the distribution of ozone pollutants in Tehran was performed. The highest concentrations of O3 are found in the southwest and parts of the central part of the city. Finally, a neural network was developed to predict the amount of ozone pollutants according to meteorological parameters. Conclusion: The results show that meteorological parameters such as temperature, velocity and direction of the wind, and precipitation are influential on O3 concentration.

POLLA: Enhancing the Local Structure Awareness in Long Sequence Spatial-temporal Modeling

The spatial-temporal modeling on long sequences is of great importance in many real-world applications. Recent studies have shown the potential of applying the self-attention mechanism to improve capturing the complex spatial-temporal dependencies. However, the lack of underlying structure information weakens its general performance on long sequence spatial-temporal problem. To overcome this limitation, we proposed a novel method, named the Proximity-aware Long Sequence Learning framework, and apply it to the spatial-temporal forecasting task. The model substitutes the canonical self-attention by leveraging the proximity-aware attention, which enhances local structure clues in building long-range dependencies with a linear approximation of attention scores. The relief adjacency matrix technique can utilize the historical global graph information for consistent proximity learning. Meanwhile, the reduced decoder allows for fast inference in a non-autoregressive manner. Extensive experiments are conducted on five large-scale datasets, which demonstrate that our method achieves state-of-the-art performance and validates the effectiveness brought by local structure information.

Spatio-Temporal Modeling of Immune Response to SARS-CoV-2 Infection

COVID-19 is a disease occurring as a result of infection by a novel coronavirus called SARS-CoV-2. Since the WHO announced COVID-19 as a global pandemic, mathematical works have taken place to simulate infection scenarios at different scales even though the majority of these models only consider the temporal dynamics of SARS-COV-2. In this paper, we present a new spatio-temporal within-host mathematical model of COVID-19, accounting for the coupled dynamics of healthy cells, infected cells, SARS-CoV-2 molecules, chemokine concentration, effector T cells, regulatory T cells, B-lymphocytes cells and antibodies. We develop a computational framework involving discretisation schemes for diffusion and chemotaxis terms using central differences and midpoint approximations within two dimensional space combined with a predict–evaluate–correct mode for time marching. Then, we numerically investigate the model performance using a list of values simulating the baseline scenario for viral infection at a cellular scale. Moreover, we explore the model sensitivity via applying certain conditions to observe the model validity in a comparison with clinical outcomes collected from recent studies. In this computational investigation, we have a numerical range of 104 to 108 for the viral load peak, which is equivalent to what has been obtained from throat swab samples for many patients.

An empirical study on temporal modeling for online action detection

Online action detection (OAD) is a practical yet challenging task, which has attracted increasing attention in recent years. A typical OAD system mainly consists of three modules: a frame-level feature extractor which is usually based on pre-trained deep Convolutional Neural Networks (CNNs), a temporal modeling module, and an action classifier. Among them, the temporal modeling module is crucial which aggregates discriminative information from historical and current features. Though many temporal modeling methods have been developed for OAD and other topics, their effects are lack of investigation on OAD fairly. This paper aims to provide an empirical study on temporal modeling for OAD including four meta types of temporal modeling methods, i.e. temporal pooling, temporal convolution, recurrent neural networks, and temporal attention, and uncover some good practices to produce a state-of-the-art OAD system. Many of them are explored in OAD for the first time, and extensively evaluated with various hyper parameters. Furthermore, based on our empirical study, we present several hybrid temporal modeling methods. Our best networks, i.e. , the hybridization of DCC, LSTM and M-NL, and the hybridization of DCC and M-NL, which outperform previously published results with sizable margins on THUMOS-14 dataset (48.6% vs. 47.2%) and TVSeries dataset (84.3% vs. 83.7%).