Wei Wang

Sea Surface Temperature (SST) is crucial for understanding upper-ocean thermal dynamics and ocean-atmosphere interactions, which have profound economic and social impacts. While data-driven models show promise in SST prediction, their black-box nature often limits interpretability and overlooks key physical processes. Recently, physics-informed neural networks have been gaining momentum but struggle with complex ocean-atmosphere dynamics due to 1) inadequate characterization of seawater movement (e.g., coastal upwelling) and 2) insufficient integration of external SST drivers (e.g., turbulent heat fluxes). To address these challenges, we propose SSTODE, a physics-informed Neural Ordinary Differential Equations (Neural ODEs) framework for SST prediction. First, we derive ODEs from fluid transport principles, incorporating both advection and diffusion to model ocean spatiotemporal dynamics. Through variational optimization, we recover a latent velocity field that explicitly governs the temporal dynamics of SST. Building upon ODE, we introduce an Energy Exchanges Integrator (EEI)-inspired by ocean heat budget equations-to account for external forcing factors. Thus, the variations in the components of these factors provide deeper insights into SST dynamics. Extensive experiments demonstrate that SSTODE achieves state-of-the-art performances in global and regional SST forecasting benchmarks. Furthermore, SSTODE visually reveals the impact of advection dynamics, thermal diffusion patterns, and diurnal heating-cooling cycles on SST evolution. These findings demonstrate the model's interpretability and physical consistency.

Satellite-derived bathymetry (SDB) based on multi-spectral imagery data has been a critical tool for large-scale water depth in shallow water regions. Traditional SDB models primarily rely on known laws relating the exponential attenuation of light with the path length it traveled. In the past few years, deep computer vision models have emerged as valuable new technologies for bathymetry measurement. However, due to the black-box nature of these deep models, they may produce bathymetry results that are inconsistent with physical laws and exhibit limited generalizability across diverse areas. In this paper, we propose a novel hybrid architecture, HybridBathNet, that integrates UNet (extracting spatial and spectral feature) with a physical bathymetry network (ensuring physical relationships). By embedding physical constraints directly into the model architecture, HybridBathNet achieves improved bathymetric inversion accuracy while maintaining consistency with established optical attenuation laws. Experimental results demonstrate that the proposed model delivers high-quality bathymetric estimations across diverse island regions. Comparative evaluations against state-of-the-art methods further validate the superior accuracy and generalization capability of HybridBathNet. The code of HybridBathNet is available at https://github.com/qiushibupt/HybridBathNet.

Social dilemmas often arise when the need for cooperation conflicts with individuals’ incentives to act in self-interest, potentially undermining collective well-being. Prior literature shows that some mechanisms, e.g., commitment, could give rise to cooperation. However, participants’ diverse propensities to cooperate may limit such mechanisms’ effectiveness. Thus, we bring individual differences in their propensities to cooperate into the reasoning of long-term social dynamics of cooperation through an agent-based modeling (ABM) approach. Our results suggest that commitment may still guarantee cooperation when individuals have different propensities to cooperate but have weaker effects, and the setups of commitment are also important. Our study highlights the importance of integrating individual preferences in analyzing collective dynamics of a population consisting of individuals of heterogeneous characteristics, thus offering implications to facilitate cooperation in real-world online social production. The model and its implementation also form the foundation for supporting decision-makers in forming decisions of facilitating cooperation via commitment mechanisms.

Cotton is a critical agricultural product and industrial raw material, playing a key role in the national economies and people's living conditions, particularly in developing countries. However, cotton picking and processing often result in the contamination with various foreign fibers, such as hair, hemp rope, plastic film, and polypropylene rope. These contaminants are difficult to remove during textile processing and tend to break into small fragments, significantly reducing the quality of cotton products and negatively impacting the cotton industry. In this paper, we present an AI-enabled hardware-software integrated system--XCotton, for identifying and removing foreign fibers. Our system has been deployed in actual cotton production environments in the multiple regions in China, Central Asia, and Africa. XCotton achieves a cleaning efficiency of 1000kg/h, representing a 43% improvement, with only 14 kWh energy consumption (63% less). Moreover, XCotton brings significant business values to its manufacturer and clients. XCotton not only enhances the quality of cotton products but also contributes to the value-adding and upgrading of the cotton industry in developing regions, supporting economic growth and improving living conditions.

Globally distributed software development has been a mainstream paradigm in developing modern software systems. We have witnessed a fast-growing population of software developers from areas where English is not a native language in the last several decades. Given that English is still the de facto working language in most global software engineering teams, we need to gain more knowledge about the experiences of developers who are non-native English speakers. We conducted an empirical study to fill this research gap. In this study, we interviewed 27 Chinese developers in commercial software development and open source global software development teams and applied Bourdieu's capital-field-habitus framework in an abductive data analysis process. Our study reveals four types of capital (language, social, symbolic, and economic) involved in their experiences and examines the interrelations among them. We found that non-native speakers' insufficient language capital played an essential role in prohibiting them from accessing and accumulating other capital, thus reproducing the sustained and systematic disadvantaged positions of non-native English speakers in GSD teams. We further discussed the theoretical and practical implications of the study.

Communication is essential in coordinating the behaviors of multiple agents. However, existing methods primarily emphasize content, timing, and partners for information sharing, often neglecting the critical aspect of integrating shared information. This gap can significantly impact agents' ability to understand and respond to complex, uncertain interactions, thus affecting overall communication efficiency. To address this issue, we introduce M2I2, a novel framework designed to enhance the agents' capabilities to assimilate and utilize received information effectively. M2I2 equips agents with advanced capabilities for masked state modeling and joint-action prediction, enriching their perception of environmental uncertainties and facilitating the anticipation of teammates' intentions. This approach ensures that agents are furnished with both comprehensive and relevant information, bolstering more informed and synergistic behaviors. Moreover, we propose a Dimensional Rational Network, innovatively trained via a meta-learning paradigm, to identify the importance of dimensional pieces of information, evaluating their contributions to decision-making and auxiliary tasks. Then, we implement an importance-based heuristic for selective information masking and sharing. This strategy optimizes the efficiency of masked state modeling and the rationale behind information sharing. We evaluate M2I2 across diverse multi-agent tasks, the results demonstrate its superior performance, efficiency, and generalization capabilities, over existing state-of-the-art methods in various complex scenarios.

Recently, unsupervised domain adaptation (UDA) for 3D object detectors has increasingly garnered attention as a method to eliminate the prohibitive costs associated with generating extensive 3D annotations, which are crucial for effective model training. Self-training (ST) has emerged as a simple and effective technique for UDA. The major issue involved in ST-UDA for 3D object detection is refining the imprecise predictions caused by domain shift and generating accurate pseudo labels as supervisory signals. This study presents a novel ST-UDA framework to generate high-quality pseudo labels by associating predictions of 3D point cloud sequences during ego-motion according to spatial and temporal consistency, named motion-associated self-training for 3D object detection (MA-ST3D). MA-ST3D maintains a global-local pathway (GLP) architecture to generate high-quality pseudo-labels by leveraging both intra-frame and inter-frame consistencies along the spatial dimension of the LiDAR’s ego-motion. It also equips two memory modules for both global and local pathways, called global memory and local memory, to suppress the temporal fluctuation of pseudo-labels during self-training iterations. In addition, a motion-aware loss is introduced to impose discriminated regulations on pseudo labels with different motion statuses, which mitigates the harmful spread of false positive pseudo labels. Finally, our method is evaluated on three representative domain adaptation tasks on authoritative 3D benchmark datasets (i.e. Waymo, Kitti, and nuScenes). MA-ST3D achieved SOTA performance on all evaluated UDA settings and even surpassed the weakly supervised DA methods on the Kitti and NuScenes object detection benchmark.

Visual interpretation is extremely important in human geography as the primary technique for geographers to use photograph data in identifying, classifying, and quantifying geographic and topological objects or regions. However, it is also time‐consuming and requires overwhelming manual effort from professional geographers. This paper describes our interdisciplinary team's efforts in integrating computer vision models with geographers' visual image interpretation process to reduce their workload in interpreting images. Focusing on the dune segmentation task, we proposed an approach called featuring a deep dune segmentation model to identify dunes and label their ranges in an automated way. By developing a tool to connect our model with ArcGIS—one of the most popular workbenches for visual interpretation, geographers can further refine the automatically generated dune segmentation on images without learning any CV or deep learning techniques. Our approach thus realized a noninvasive change to geographers' visual interpretation routines, reducing their manual efforts while incurring minimal interruptions to their work routines and tools they are familiar with. Deployment with a leading Chinese geography research institution demonstrated the potential of in supporting geographers in researching and solving drylands desertification.

Dune segmentation is a fundamental task for geographers’ efforts in tackling the global desertification problems. Geographers often use visual interpretation to segment dunes, thus developing valuable insights into the ever-changing conditions that shape dunes. However, visual interpretation is performed manually, which is time-consuming and requires significant manual effort from professional geographers. Recently, Segment Anything Model (SAM) has been introduced as a universal pre-trained framework for image segmentation tasks based on 1 billion masks and 11 million images. While SAM demonstrates impressive zero-shot generalization in conventional segmentation tasks, its effectiveness diminishes when applied to dune segmentation, primarily due to the absence of geography-specific knowledge. Therefore, employing the large foundational model SAM for dune-related tasks with minimal annotations is of significant practical value. In this paper, instead of fine-tuning all parameters of the SAM network, we propose the Dune SAM Adapter model (DuneSA), which incorporates domain-specific knowledge about dunes into SAM using a lightweight yet effective adaptation technique. Specifically, in the encoder part of SAM, we integrate a series of adapters, including the ScaleAdapter within and the FeatureAdapter between the Vision Transformer (ViT) blocks, to achieve performance that meets geographers’ standards. We conduct comprehensive evaluation experiments in Mu Us Sandy Land with a leading Chinese geography research institution. The results show that by integrating task-specific knowledge with general knowledge learned by the large model, our approach can assist geographers in researching and addressing desertification.