报告题目：Influence Propagation Based on Strong Dominance in a Hybrid Linear Threshold SIS Model
时 间：2023-08-14 (星期一) 15:00 -18:00
地 点：科研楼18号楼1102
主 办：数学与统计学院, 分析数学及应用教育部重点实验室、福建省应用数学中心（福建师范大学）
参加对象：感兴趣的老师和研究生
报告摘要：Influence maximization (IM) is a fundamental algorithmic problem that aims to find a set of $k$ users from a social network, referred to as a seed set, to yield the maximum influence spread results. In this paper, we employ a strong dominance strategy to sieve the initial seed set to deal with the evaluation of influence propagation, inspired by the fact that strong dominance relations have been shown to enclose more outward influence walks in ecological competition. To evaluate the performance of influence propagation, we adopt a hybrid linear threshold SIS model to scrutinize the regularity and variation of its propagation behavior. Specifically, unlike the traditional linear threshold model, we allow individuals to divert their state between susceptible and infected. We conduct propagation experiments on four real networks, in which the influence factor of each individual in the social network is assigned according to the Pareto principle, and the threshold for receiving infection is set by the normal distribution. Therefore, we performed various differential analyses on the infected proportion, including differences in the individual degree distribution, adjustments of the number of seeds, the lengths of the recovery time (i.e., the period for transforming infection into susceptibility), and the influence delay under diverse conditions.
报告人简介：张肇明教授是台北商业大学特聘教授，2011至2013年担任资讯与决策科学研究所所长，2014至2015年担任台北商业大学管理学院院长，同时兼任“演算法与计算理论学会”理事一职。到目前为止，已经连续21年（共15次）获得台湾科技专题研究项目主持人。在学术研究上，在IEEE/ACM ToN, IEEE TC, IEEE TPDS等发表110余篇期刊论文与100余篇会议论文。目前，主要的研究领域包括算法设计与分析，图论，并行与分布式计算等主题。