Federated learning has wide applications in the medical field. It enables knowledge sharing among different healthcare institutes while protecting patients’ privacy. However, existing federated learning systems are typically centralized, requiring clients to upload client-specific knowledge to a central server for aggregation. This centralized approach would integrate the knowledge from each client into a centralized server, and the knowledge would be already undermined during the centralized integration before it reaches back to each client. Besides, the centralized approach also creates a dependency on the central server, which may affect training stability if the server malfunctions or connections are unstable. To address these issues, we propose a decentralized federated learning framework named dFLMoE. In our framework, clients directly exchange lightweight head models with each other. After exchanging, each client treats both local and received head models as individual experts, and utilizes a client-specific Mixture of Experts (MoE) approach to make collective decisions. This design not only reduces the knowledge damage with client-specific aggregations but also removes the dependency on the central server to enhance the robustness of the framework. We validate our framework on multiple medical tasks, demonstrating that our method evidently outperforms state-of-the-art approaches under both model homogeneity and heterogeneity settings.

The increase in highway mileage and lifespan is driving up the demand for road maintenance. With most research focusing on corrective maintenance, remedial maintenance(such as sealing and patching) optimization is understudied. Oriented toward remedial maintenance, data-driven models often fall short due to difficulty in establishing and implementing the model under complex road conditions, while the experts’ decision lacks consistency amidst multifaceted factors. To address this gap, this paper proposes a fine-grained maintenance decision model that combines data-driven methods with expert knowledge through Reinforcement Learning from Expert Feedback (RLEF). The experts’ experience introduced in decision-making model could improve the engineering application ability of decisions. The research uses a pavement performance prediction model as the environment and applies reinforcement learning to optimize strategies in the decision model. Additionally, the model integrates multidimensional expert feedback into reward functions to better understand ambiguous decision rules. Real-world data validation demonstrates that the RLEF model can adapt to engineering scenarios and applications better as well as achieve superior cost-effectiveness.

Pavement performance prediction is the foundation of maintenance decisions, which is the key problem of infrastructure management. Most prediction methods focus on section-based and annual deterioration on pavement, while it is hardly supporting daily and preventive maintenance plans. To fill in gaps in pavement forecasting for refined maintenance, this paper introduces a prediction model which is fine-grained on both temporal and spatial scales. Due to the coupling effects and action delay of multiple environmental factors, it is difficult to fathom and model the detailed deterioration process of pavement. Another problem is there are rare refined pavement datasets opening to public for research. Therefore, we establish a high-frequency and real-world pavement dataset and causal discovery is brought in to explicate the inner mechanism of the process. The proposed model first applies Partial Mutual Information from Mixed Embedding (PMIME) method for causal discovery, obtaining a causal graph and impact delays between factors and pavement performance. Based on this, we use an advanced pavement performance prediction model called Causal-Temporal Graph Convolution Network (CTGCN), combining the Graph Convolution Networks (GCNs) and the Long Short-Term Memory models (LSTMs) to capture causal features and temporal features simultaneously. We validate CTGCN model using collected datasets with two predictive time lengths. The experimental results prove that CTGCN model has better performance in both prediction accuracy and robustness than the state-of-art baseline. Dataset and more information are available at https://github.com/wowocai/CTGCN-dataset.

Recently most handwritten mathematical expression recognition methods adopt the attention-based encoder-decoder framework, which generates LaTeX sequences from given images. However, the accuracy of the attention mechanism limits the performance of HMER models. Lacking global context information in the decoding process is also a challenge for HMER. Some methods adopt symbol-level counting to localize symbols for improving the model performance, while these methods cannot work well. In this paper, we propose a method named SLAN, shorted for a Symbol Location-Aware Network, to solve the HMER problem. Specifically, we propose an advanced relation-level counting method to detect symbols in the image. We solve the lacking global context problem with a new global context-aware decoder. For improving the accuracy of attention, we design a novel attention alignment loss function by the dynamic programming algorithm, which can learn attention alignment directly without pixel-level labels. We conducted extensive experiments on the CROHME dataset to demonstrate the effectiveness of each part of SLAN and achieved state-of-the-art performance.

Program understanding is a fundamental task in program language processing.Despite the success, existing works fail to take human behaviors as reference in understanding programs.In this paper, we consider human behaviors and propose the PGNN-EK model that consists of two main components.On the one hand, inspired by the "divide-and-conquer" reading behaviors of humans, we present a partitioningbased graph neural network model PGNN on the upgraded AST of codes.On the other hand, to characterize human behaviors of resorting to other resources to help code comprehension, we transform raw codes with external knowledge and apply pre-training techniques for information extraction.Finally, we combine the two embeddings generated from the two components to output code embeddings.We conduct extensive experiments to show the superior performance of PGNN-EK on the code summarization and code clone detection tasks.In particular, to show the generalization ability of our model, we release a new dataset that is more challenging for code clone detection and could advance the development of the community.

Author disambiguation arises when different authors share the same name, which is a critical task in digital libraries, such as DBLP, CiteULike, CiteSeerX, etc. While the state-of-the-art methods have developed various paper embedding-based methods performing in a top-down manner, they primarily focus on the ego-network of a target name and overlook the low-quality collaborative relations existed in the ego-network. Thus, these methods can be suboptimal for disambiguating authors.In this paper, we model the author disambiguation as a collaboration network reconstruction problem, and propose an incremental and unsupervised author disambiguation method, namely IUAD, which performs in a bottom-up manner. Initially, we build a stable collaboration network based on stable collaborative relations. To further improve the recall, we build a probabilistic generative model to reconstruct the complete collaboration network. In addition, for newly published papers, we can incrementally judge who publish them via only computing the posterior probabilities. We have conducted extensive experiments on a large-scale DBLP dataset to evaluate IUAD. The experimental results demonstrate that IUAD not only achieves the promising performance, but also outperforms comparable baselines significantly. Codes are available at https://github.com/papergitgit/IUAD.