Search

This paper focuses on a specific family of classifiers called nonparallel support vector classifiers (NPSVCs). Different from typical classifiers, the training of an NPSVC involves the minimization of multiple objectives, resulting in the potential concerns of feature suboptimality and class dependency. Consequently, no effective learning scheme has been established to improve NPSVCs' performance through representation learning, especially deep learning. To break this bottleneck, we develop NPSVC++ based on multi-objective optimization, enabling the end-to-end learning of NPSVC and its features. By pursuing Pareto optimality, NPSVC++ theoretically ensures feature optimality across classes, hence effectively overcoming the two issues above. A general learning procedure via duality optimization is proposed, based on which we provide two applicable instances, K-NPSVC++ and D-NPSVC++. The experiments show their superiority over the existing methods and verify the efficacy of NPSVC++.

Vehicular Metaverses represent emerging paradigms arising from the convergence of vehicle road cooperation, Metaverse, and augmented intelligence of things. Users engaging with Vehicular Metaverses (VMUs) gain entry by consistently updating their Vehicular Twins (VTs), which are deployed on RoadSide Units (RSUs) in proximity. The constrained RSU coverage and the consistently moving vehicles necessitate the continuous migration of VTs between RSUs through vehicle road cooperation, ensuring uninterrupted immersion services for VMUs. Nevertheless, the VT migration process faces challenges in obtaining adequate bandwidth resources from RSUs for timely migration, posing a resource trading problem among RSUs. In this paper, we tackle this challenge by formulating a game-theoretic incentive mechanism with multi-leader multi-follower, incorporating insights from social-awareness and queueing theory to optimize VT migration. To validate the existence and uniqueness of the Stackelberg Equilibrium, we apply the backward induction method. Theoretical solutions for this equilibrium are then obtained through the Alternating Direction Method of Multipliers (ADMM) algorithm. Moreover, owing to incomplete information caused by the requirements for privacy protection, we proposed a multi-agent deep reinforcement learning algorithm named MALPPO. MALPPO facilitates learning the Stackelberg Equilibrium without requiring private information from others, relying solely on past experiences. Comprehensive experimental results demonstrate that our MALPPO-based incentive mechanism outperforms baseline approaches significantly, showcasing rapid convergence and achieving the highest reward.

Vehicular metaverses are an emerging paradigm that integrates extended reality technologies and real-time sensing data to bridge the physical space and digital spaces for intelligent transportation, providing immersive experiences for Vehicular Metaverse Users (VMUs). VMUs access the vehicular metaverse by continuously updating Vehicular Twins (VTs) deployed on nearby RoadSide Units (RSUs). Due to the limited RSU coverage, VTs need to be continuously online migrated between RSUs to ensure seamless immersion and interactions for VMUs with the nature of mobility. However, the VT migration process requires sufficient bandwidth resources from RSUs to enable online and fast migration, leading to a resource trading problem between RSUs and VMUs. To this end, we propose a learning-based incentive mechanism for migration task freshness-aware VT migration in vehicular metaverses. To quantify the freshness of the VT migration task, we first propose a new metric named Age of Twin Migration (AoTM), which measures the time elapsed of completing the VT migration task. Then, we propose an AoTM-based Stackelberg model, where RSUs act as the leader and VMUs act as followers. Due to incomplete information between RSUs and VMUs caused by privacy and security concerns, we utilize deep reinforcement learning to learn the equilibrium of the Stackelberg game. Numerical results demonstrate the effectiveness of our proposed learning-based incentive mechanism for vehicular metaverses., Comment: 2023 IEEE 43rd International Conference on Distributed Computing Systems Workshops, ICDCSW 2023

As the fusion of automotive industry and metaverse, vehicular metaverses establish a bridge between the physical space and virtual space, providing intelligent transportation services through the integration of various technologies, such as extended reality and real-time rendering technologies, to offer immersive metaverse services for Vehicular Metaverse Users (VMUs). In vehicular metaverses, VMUs update vehicle twins (VTs) deployed in RoadSide Units (RSUs) to obtain metaverse services. However, due to the mobility of vehicles and the limited service coverage of RSUs, VT migration is necessary to ensure continuous immersive experiences for VMUs. This process requires RSUs to contribute resources for enabling efficient migration, which leads to a resource trading problem between RSUs and VMUs. Moreover, a single RSU cannot support large-scale VT migration. To this end, we propose a blockchain-assisted game approach framework for reliable VT migration in vehicular metaverses. Based on the subject logic model, we first calculate the reputation values of RSUs considering the freshness of interaction between RSUs and VMUs. Then, a coalition game based on the reputation values of RSUs is formulated, and RSU coalitions are formed to jointly provide bandwidth resources for reliable and large-scale VT migration. Subsequently, the RSU coalition with the highest utility is selected. Finally, to incentivize VMUs to participate in VT migration, we propose a Stackelberg model between the selected coalition and VMUs. Numerical results demonstrate the reliability and effectiveness of the proposed schemes., Comment: Transactions on Emerging Telecommunications Technologies (ISSN: 2161-3915)