Artificial Immune System for Proportionality Assessment in Military Operations

Abstract

This article presents an Artificial Immune System (AIS) algorithm for conducting the proportionality assessment in military operations that integrates an in-depth evaluation in relation to the detection of disproportionate collateral damage in military operations under two modelling settings, as follows defined. In the first modelling setting, the collateral damage only includes physical harm, and in the second modelling setting, the collateral damage includes both physical and psychological harm. In this context, various algorithms were implemented and compared. By employing a Clonal Selection algorithm, the possible scenario‐feature combinations are treated as antigens and evolve a detector repertoire via affinity‐based cloning, mutation, and replacement and aspects such as detection rate, false‐positive rate, and detector diversity are tracked over generations. By employing Negative Selection and Immune Network paradigms, it can be seen that in Case 2 superior disproportional detection is achieved and through a more in-depth network analysis the core communities that underpin disproportionate judgments are analysed. Hence, the AIS models developed demonstrate that incorporating psychological considerations not only improves detection performance, but also shapes the structural properties of immune repertoires, providing a novel computational perspective on the proportionality assessment execution in military operations.