Skip to main content



Studying the Game Theory of Criminal Street Gangs to Predict Behavior of Adversaries

The USC Center for Artificial Intelligence in Society is proposing a project called Spatio-Temporal Game Theory & Real-Time Machine Learning for Advesarial Groups. This project would combine smart-phone based game theory games and gang related police data to model behavior between adversaries. These predictions can be applied beyond gang relations to include adversarial relationships among extremist groups, like ISIS. The project stipulates that individuals in extremist group networks may form weak local bridges to other extremist groups, but that mainly there are strong and violent rivalries between groups. So while this project focuses mainly on criminal gangs in the United States, it seeks to predict adversary trajectories in other arenas, like global politics. Their methodology uses ML and algorithmic game theory.

This project is very much related to INFO 2040 for its intersection of network principles and real life social issues. It discusses Game Theory principles as related to the choices made by gang members to cooperate or not. Circumstances of the “game” are manipulated – whether the partner is a cooperator or not, whether the game is occurring within gang territory or not, and speed – to analyze their effects on decisions and outcomes. This project is also related to concept of structural balance since the relationships between gang members and members of opponent gangs produce rivalries. A structurally balanced triangle is formed between two members of the same gang and one member of a rival gang; the two have a common enemy in the one (2 negative edges, 1 positive edge). Overall, the principles of game theory, local bridges, and structural balance are embedded within this project.

 

https://www.cais.usc.edu/projects/gametheory/

Comments

Leave a Reply

Blogging Calendar

September 2018
M T W T F S S
 12
3456789
10111213141516
17181920212223
24252627282930

Archives