Social learning under inferential attacks

TL;DR

This paper investigates how malicious agents can manipulate belief updates in social learning networks through inferential attacks, especially when agents have limited knowledge of the network, and characterizes conditions for successful deception.

Contribution

It introduces a framework for understanding inferential attacks in social learning, detailing strategies malicious agents can use and conditions leading to network misledness.

Findings

Conditions for successful inferential attacks identified

Strategies for malicious agents to manipulate beliefs developed

Analysis of attack effectiveness with limited network information

Abstract

A common assumption in the social learning literature is that agents exchange information in an unselfish manner. In this work, we consider the scenario where a subset of agents aims at driving the network beliefs to the wrong hypothesis. The adversaries are unaware of the true hypothesis. However, they will "blend in" by behaving similarly to the other agents and will manipulate the likelihood functions used in the belief update process to launch inferential attacks. We will characterize the conditions under which the network is misled. Then, we will explain that it is possible for such attacks to succeed by showing that strategies exist that can be adopted by the malicious agents for this purpose. We examine both situations in which the agents have minimal or no information about the network model.