Publication Type

Conference Proceeding Article

Version

acceptedVersion

Publication Date

5-2016

Abstract

Wildlife poaching presents a serious extinction threat to many animalspecies. Agencies (“defenders”) focused on protecting suchanimals need tools that help analyze, model and predict poacheractivities, so they can more effectively combat such poaching; suchtools could also assist in planning effective defender patrols, buildingon the previous security games research.To that end, we have built a new predictive anti-poaching tool,CAPTURE (Comprehensive Anti-Poaching tool with Temporaland observation Uncertainty REasoning). CAPTURE providesfour main contributions. First, CAPTURE’s modeling of poachersprovides significant advances over previous models from behavioralgame theory and conservation biology. This accounts for:(i) the defender’s imperfect detection of poaching signs; (ii) complextemporal dependencies in the poacher’s behaviors; (iii) lackof knowledge of numbers of poachers. Second, we provide twonew heuristics: parameter separation and target abstraction to reducethe computational complexity in learning the poacher models.Third, we present a new game-theoretic algorithm for computingthe defender’s optimal patrolling given the complex poachermodel. Finally, we present detailed models and analysis of realworldpoaching data collected over 12 years in Queen ElizabethNational Park in Uganda to evaluate our new model’s predictionaccuracy. This paper thus presents the largest dataset of real-worlddefender-adversary interactions analyzed in the security games literature.CAPTURE will be tested in Uganda in early 2016.

Discipline

Databases and Information Systems

Research Areas

Data Science and Engineering

Publication

Proceedings of 15th International Conference on Autonomous Agents and Multiagent Systems (AAMAS)

City or Country

Singapore

Download

Share

COinS