Building agents with adaptive behavior in cooperative tasks stands as a paramount goal in the realm of multi-agent systems. Current approaches to developing cooperative agents rely primarily on learning-based methods, whose policy generalization depends heavily on the diversity of teammates they interact with during the training phase. Such reliance, however, constrains the agents' capacity for strategic adaptation when cooperating with unfamiliar teammates, which becomes a significant challenge in zero-shot coordination scenarios. To address this challenge, we propose ProAgent, a novel framework that harnesses large language models (LLMs) to create proactive agents capable of dynamically adapting their behavior to enhance cooperation with teammates. ProAgent can analyze the present state, and infer the intentions of teammates from observations. It then updates its beliefs in alignment with the teammates' subsequent actual behaviors. Moreover, ProAgent exhibits a high degree of modularity and interpretability, making it easily integrated into various of coordination scenarios. Experimental evaluations conducted within the Overcooked-AI environment unveil the remarkable performance superiority of ProAgent, outperforming five methods based on self-play and population-based training when cooperating with AI agents. Furthermore, in partnered with human proxy models, its performance exhibits an average improvement exceeding 10% compared to the current state-of-the-art method. For more information about our project, please visit https://pku-proagent.github.io.

In offline reinforcement learning (RL), the performance of the learned policy highly depends on the quality of offline datasets. However, the offline dataset contains very limited optimal trajectories in many cases. This poses a challenge for offline RL algorithms, as agents must acquire the ability to transit to high-reward regions. To address this issue, we introduce Diffusionbased Trajectory Stitching (DiffStitch), a novel diffusion-based data augmentation pipeline that systematically generates stitching transitions between trajectories. DiffStitch effectively connects low-reward trajectories with high-reward trajectories, forming globally optimal trajectories and thereby mitigating the challenges faced by offline RL algorithms in learning trajectory stitching. Empirical experiments conducted on D4RL datasets demonstrate the effectiveness of our pipeline across RL methodologies. Notably, DiffStitch demonstrates substantial enhancements in the performance of one-step methods(IQL), imitation learning methods(TD3+BC) and trajectory optimization methods(DT). Our code is publicly available at https://github.com/guangheli12/DiffStitch

The outbreak of Hand-Foot-Mouth Disease (HFMD) has brought a serious threat to people's health. There is no research considering both the natural and social environment to study the outbreak and propagation of HFMD quantitatively by now. In this study, we acquired environmental impact factors of HFMD using remote sensing and GIS. Then we analyzed the relationship between density of HFMD cases and environmental impact factors for year 2009 in Shenzhen. It was found that density of HFMD cases was significantly associated with population density, hospital density, road density, annual average of aerosol optical depth, and annual average of normalized difference vegetation index. A linear regression model was constructed with the R2 value of 0.677 using environmental impact factors, and both hospital density and annual average of aerosol optical depth were included in the model.

Wenchuan in Sichuan Province, China, was hit by an 8.0 magnitude earthquake on 12 May 2008. Thousands of buildings were destroyed and thousands of people were hurt. History tells us that after earthquakes, floods and other natural disaster, infectious diseases often outbreaks. In this study, with the help of spatial information technologies including remote sensing and geography information system, we acquired environmental factors including population, earthquake intensity, route distance, direct distance, Normalized Difference Vegetation Index, Normalized Difference Water Index, and Digital Elevation Model for each township in Anxian after earthquake. Then we constructed a Back Propagation Neural Network to train environmental factors and incidence of communicable diseases, and predicted the incidence of communicable diseases. The results showed that incidence of communicable diseases after earthquake could be predicted using remote sensing to some extent.