Neural Influence Estimator: Towards Real-time Solutions to Influence Blocking Maximization

TL;DR

The paper tackles the IBM problem on large social networks under tight real-time constraints, proposing NIE, a neural surrogate trained offline to predict the blocked influence and replace expensive Monte Carlo simulations. NIE is coupled with an IBM optimization routine (NIE-CELF) to deliver real-time solutions with strong scalability, achieving up to four orders of magnitude speedups over MCSs-based solvers while handling networks with hundreds of thousands of nodes. The approach relies on a multi-faceted feature extraction that captures both local/topological properties of seed sets and their inter-relationships, enabling accurate predictions of $f(S_t|S_f)$ via a lightweight MLP. Empirical results across five networks and 25 IBM problems demonstrate substantial runtime gains and competitive optimization quality within a one-minute budget, supporting practical deployment for misinformation containment and offering a pathway to extendable surrogate-based optimization in graph-based problems with costly evaluations.

Abstract

Real-time solutions to the influence blocking maximization (IBM) problems are crucial for promptly containing the spread of misinformation. However, achieving this goal is non-trivial, mainly because assessing the blocked influence of an IBM problem solution typically requires plenty of expensive Monte Carlo simulations (MCSs). This work presents a novel approach that enables solving IBM problems with hundreds of thousands of nodes and edges in seconds. The key idea is to construct a fast-to-evaluate surrogate model called neural influence estimator (NIE) offline as a substitute for the time-intensive MCSs, and then combine it with optimization algorithms to address IBM problems online. To this end, a learning problem is formulated to build the NIE that takes the false-and-true information instance as input, extracts features describing the topology and inter-relationship between two seed sets, and predicts the blocked influence. A well-trained NIE can generalize across different IBM problems given a social network, and can be readily combined with existing IBM optimization algorithms. The experiments on 25 IBM problems with up to millions of edges show that the NIE-based optimization method can be up to four orders of magnitude faster than MCSs-based optimization method to achieve the same optimization quality. Moreover, given a one-minute limit, the NIE-based method can solve IBM problems with up to hundreds of thousands of nodes, which is at least one order of magnitude larger than what can be solved by existing methods.

Figures (5)

• Figure 1: The architecture of NIE. The Inter-relationship Feature Extractor includes function $p(S_f, S_t)$ that quantifies the inter-relationship between $S_f$ and $S_t$. The Topological Feature Extractor extracts the topological patterns of $S_f$ and $S_t$, including the neighborhood feature $d$, the location feature $b$, and the structure feature $c$. The MLP predicts the blocked influence based on the extracted features.
• Figure 2: Curves of the optimization quality (i.e. the blocked influence) vs. runtime of the optimization methods on five test problems, within a time budget of 48 hours. The result of ACO-GE is shown by one point because it is a rank-based algorithm rather than an iterative optimization algorithm.
• Figure 3: Feature importance analysis for NIE. $d_f$, $b_f$, and $c_f$ represents the neighborhood feature, the location feature, and the structure feature of $S_f$, respectively. $d_t$, $b_t$, and $c_t$ represents the neighborhood feature, the location feature, and the structure feature of $S_t$, respectively. $p$ denotes the inter-relationship feature between $S_f$ and $S_t$.
• Figure 4: The optimization quality (i.e. the blocked influence) achieved by NIE-CELF under different values of $H$.
• Figure 5: Performance of MLP under different hyperparameters values. The triangle represents the MSE of the baseline configuration, and circles represent the MSE of the modified configurations.

Theorems & Definitions (2)

• Definition 1
• Definition 2