Entity Resolution (ER) is vital for data integration and knowledge graph construction. Despite the advancements made by deep learning (DL) methods using pre-trained language models (
Geometric graphs are a special kind of graph with geometric features, which are vital to model many scientific problems. Unlike generic graphs, geometric graphs often exhibit physical symmetries of translations, rotations, and reflections, making them ineffectively processed by current Graph Neural Networks (GNNs). To address this issue, researchers proposed a variety of geometric GNNs equipped with invariant/equivariant properties to better characterize the geometry and topology of geometric graphs. Given the current progress in this field, it is imperative to conduct a comprehensive survey of data structures, models, and applications related to geometric GNNs. In this paper, based on the necessary but concise mathematical preliminaries, we formalize geometric graph as the data structure, on top of which we provide a unified view of existing models from the geometric message passing perspective. Additionally, we summarize the applications as well as the related datasets to facilitate later research for methodology development and experimental evaluation. We also discuss the challenges and future potential directions of geometric GNNs at the end of this survey.
This paper presents a novel generative model, Collaborative Competitive Agents (CCA), which leverages the capabilities of multiple Large Language Models (LLMs) based agents to execute complex tasks. Drawing inspiration from Generative Adversarial Networks (GANs), the CCA system employs two equal-status generator agents and a discriminator agent. The generators independently process user instructions and generate results, while the discriminator evaluates the outputs, and provides feedback for the generator agents to further reflect and improve the generation results. Unlike the previous generative model, our system can obtain the intermediate steps of generation. This allows each generator agent to learn from other successful executions due to its transparency, enabling a collaborative competition that enhances the quality and robustness of the system’s results. The primary focus of this study is image editing, demonstrating the CCA’s ability to handle intricate instructions robustly. The paper’s main contributions include the introduction of a multi-agent-based generative model with controllable intermediate steps and iterative optimization, a detailed examination of agent relationships, and comprehensive experiments on image editing.
Symmetry is prevalent in multi-agent systems. The presence of symmetry, coupled with the misuse of absolute coordinate systems, often leads to a large amount of redundant representation space, significantly increasing the search space for learning policies and reducing learning efficiency. Effectively utilizing symmetry and extracting symmetry-invariant representations can significantly enhance multi-agent systems’ learning efficiency and overall performance by compressing the model’s hypothesis space and improving sample efficiency. The issue of rotational symmetry in multi-agent reinforcement learning has received little attention in previous research and is the primary focus of this paper. To address this issue, we propose a rotation-invariant network architecture for continuous action space tasks. This architecture utilizes relative coordinates between agents, eliminating dependence on absolute coordinate systems, and employs a hypernetwork to enhance the model’s fitting capability, enabling it to model MDPs with more complex dynamics. It can be used for both predicting actions and evaluating action values/utilities. In benchmark tasks, experimental results validate the impact of rotational symmetry on multi-agent decision systems and demonstrate the effectiveness of our method. The code of RDHNet has been available at the website of github.com/wang88256187/RDHNet.
Educational Cognitive Diagnosis (CD) aims to provide students’ mastery levels on different concepts. One common observation is that students often conduct many exercises but engage with a small subset of concepts, leading to a sparsity barrier. Current CD models mostly adopt mastery levels on all concepts as student modeling, overlooking the sparsity barrier. If a student does not interact with all concepts, we can not ensure that each dimension of mastery levels on concepts can be well-trained. In this paper, we propose a novel Enhancing Student Representations in Cognitive Diagnosis (ESR-CD), which combines application abilities and comprehension degrees for mastery levels on concepts. To model application ability, we propose a sparsity-based mask module that solely depends on the dense student-concept entries. Simultaneously, to further enhance comprehension degrees, we propose two layers: a matrix factorization layer and a relation refinement layer. Extensive experiments on two real-world datasets demonstrate the effectiveness of ESR-CD.
Based on the foundation of Large Language Models (LLMs), Multilingual LLMs (MLLMs) have been developed to address the challenges faced in multilingual natural language processing, hoping to achieve knowledge transfer from high-resource languages to low-resource languages. However, significant limitations and challenges still exist, such as language imbalance, multilingual alignment, and inherent bias. In this paper, we aim to provide a comprehensive analysis of MLLMs, delving deeply into discussions surrounding these critical issues. First of all, we start by presenting an overview of MLLMs, covering their evolutions, key techniques, and multilingual capacities. Secondly, we explore the multilingual training corpora of MLLMs and the multilingual datasets oriented for downstream tasks that are crucial to enhance the cross-lingual capability of MLLMs. Thirdly, we survey the state-of-the-art studies of multilingual representations and investigate whether the current MLLMs can learn a universal language representation. Fourthly, we discuss bias on MLLMs, including its categories, evaluation metrics, and debiasing techniques. Finally, we discuss existing challenges and point out promising research directions of MLLMs.
Mobile CrowdSensing (MCS) has become a powerful sensing paradigm for information collection recently. As sensing becomes more complicated, it is beneficial to deploy edge servers between users and the cloud center with a so-called mobile edge computing. Instead of directly offloading the sensing data to the cloud center, mobile users offload the sensing data to the edge servers. Then, the edge server processes and transmits the data to the cloud center in a distributed and parallel manner. It’s however critically important to balance cost, such as energy consumption, and the stability of the queues on both mobile users and edge servers. Therefore, to minimize the data offloading cost while maintaining system stability, we should carefully design the sensing data offloading strategy for edge-based crowdsensing. To this end, we formulate a double-queue Lyapunov optimization problem and propose a sensing data offloading strategy. We analyze the upper bounds of the total offloading cost and queue backlog. We further formulate the heterogeneous sensing data problem as the minimum weight bipartite graph matching problem and develop an approach that is based on Kuhn-Munkres algorithm. Finally, we conduct simulations based on three mobility sets. Simulation results show that the proposed techniques outperform several state-of-art algorithms in overall cost, system stability, and other performance metrics.
Software Defined Networking (SDN) offers Traffic Engineering (TE) great flexibility by decoupling the control and data plane. As network services become more diverse, the single-domain control architecture is no longer sufficient to meet scalability requirements, and the multi-domain and multi-level distributed control architecture is gaining popularity. However, traffic engineering across multiple domains poses challenges, particularly when each administrative domain is unwilling to disclose its network topology and resource information due to privacy concerns. To address this issue, this paper adopts the concept of differential privacy and perturbs the domain information to achieve bandwidth indistinguishable TE. Unfortunately, perturbations may decrease the accuracy of the TE algorithm’s resource allocation, negatively affecting performance. To mitigate this problem, we propose BI-TE, which utilizes a GNN-based bandwidth utilization prediction model to assist the controller in selecting the optimal forwarding path, thereby enhancing TE efficiency. Experimental results demonstrate that compared to abstraction-based hierarchical TE, BI-TE can reduce the processing time by nearly 24.35% while ensuring network bandwidth utilization close to 90%. Additionally, the fairness of allocation is also guaranteed.
