Search

Abstract The influence maximization problem is one of the most fundamental topics in social networks. However, most existing studies have focused on non-attributed networks, neglecting the consideration of users’ properties during information propagation. Additionally, specific scenarios may involve external queries that target a particular subset of users, which has not been adequately addressed in prior research. To address these limitations, this study first formulates the customized influence maximization (CIM) problem in the context of attributed social networks. The node score and influence probability are derived by fully considering the user’s attributes and the external queries. Then, we develop two algorithms to identify a group of most influential nodes in CIM. The first is a heuristic algorithm based on discounted degree, which is able to find relatively high-quality solutions in a short time. The second is a meta-heuristic algorithm, which makes several adjustments to the original ant colony algorithm to make it efficient to the CIM problem. Specifically, multiple CIM-related heuristics are derived, and a heuristic adaptation strategy is designed to automatically assign the heuristic information to ants according to the search environments and stages. Extensive experiments show the promising performance of our proposed algorithms in terms of accuracy, efficiency, and robustness.

Abstract Document clustering has long been an important research direction in intelligent system. When being applied to process Chinese documents, new challenges were posted since it is infeasible to directly split the Chinese documents using the whitespace character. Moreover, many Chinese document clustering algorithms require prior knowledge of the cluster number, which is impractical to know in real-world applications. Considering these problems, we propose a general Chinese document clustering framework, where the main clustering task is fulfilled with an adaptive encoding-based evolutionary approach. Specifically, the adaptive encoding scheme is proposed to automatically learn the cluster number, and novel crossover and mutation operators are designed to fit this scheme. In addition, a single step of K-means is incorporated to conduct a joint global and local search, enhancing the overall exploitation ability. The experiments on benchmark datasets demonstrate the superiority of the proposed method in both the efficiency and the clustering precision.

Abstract Traffic Signal Control (TSC) is a fundamental task in modern intelligent transport systems. TSC is often formulated as a bi‐level optimization problem, comprised by the signal timing at the upper level and the Dynamic User Equilibrium (DUE) traffic assignment at the lower level. Since DUE is non‐convex, existing methods either formulate approximation models or adopt traffic simulators. However, approximation models may oversimplify the practical situations, while traffic simulators are usually time‐consuming. This paper formulates a vehicle‐based DUE (vDUE) model and proposes an agile method that can simultaneously maintain the computational simplicity and the traffic dynamics for the traffic assignment. Further, an agile TSC system is built by combining the vDUE at the lower level for the traffic assignment with an adaptive differential evolution algorithm at the upper level for the signal timing optimization. To enhance the effectiveness of optimization, the TSC problem formulation is also improved to make it better characterize the practical requirements. In the experiments undertaken, comparisons of different TSC methods are carried out on both real‐world and synthetic transportation networks. The experimental results validate the effectiveness of the proposed agile TSC system in various traffic situations.

Demand side management is an efficient way of reducing the grid fluctuation and enhancing the net profit in the smart grid. As the increasing complexity of the real-world applications, e.g., NP-hard and non-polynomial problems, a distributed coevolution algorithm is proposed to cope with this challenge. In the proposed algorithm, the management system is modeled as a hierarchical structure. The user side locates at the lowest level and upper levels represent the system operators. The system operators manage users' energy consumption behavior through the real-time pricing strategy. The genetic algorithm and particle swarm optimization are modified to play the role of the smart scheduler and the smart pricing generator. The end users conduct appliance commitment using the smart scheduler which considers the users' comfort level and the electricity cost. The decision of the lower level depends directly on the real-time pricing strategy released by its adjacent upper level. The real-time pricing is developed by the smart pricing generator which considers the net profit and the grid fluctuation. Three types of experiments are conducted on a distributed platform to investigate and ascertain the performance of the proposed algorithm. For the operator side, the experimental results indicate that the proposed algorithm improves the grid fluctuation, and enhances the net profit. As for the user side, it improves the comfort level and achieves budget saving.

Multi-label learning plays a critical role in the areas of data mining, multimedia, and machine learning. Although many multi-label approaches have been proposed, few of them have considered to de-emphasize the effect of noisy features in the learning process. To address this issue, this paper designs a new method named representative multi-label learning algorithm. Instead of considering all features, the proposed algorithm focuses only on the representative ones, via incorporating an affinity propagation algorithm, kernel formulation, and a multi-label support vector machine into the learning framework. Specifically, it first adopts an affinity propagation algorithm to select a set of representative features and capture the relationships among features. Then, the algorithm constructs the representative kernel functions to measure the similarity between data instances. Finally, a multi-label support vector machine is applied to solve the learning problem. Based on the representative multi-label learning algorithm, we further design a representative multi-label learning ensemble framework to improve the accuracy, stableness, and robustness. Experimental results show that the proposed algorithm works well on most of the datasets and outperforms the compared multi-label learning approaches.