Stay Ahead of Poachers: Illegal Wildlife Poaching Prediction and Patrol Planning Under Uncertainty with Field Test Evaluations.

摘要: Illegal wildlife poaching threatens ecosystems and drives endangered species toward extinction. However, efforts for protection are constrained by the limited resources of law enforcement agencies. To help combat poaching, Protection Assistant Wildlife Security (PAWS) is a machine learning pipeline that has been developed as data-driven approach to identify areas at high risk throughout protected compute optimal patrol routes. In this paper, we take an end-to-end data-to-deployment anti-poaching. doing so, address challenges including extreme class imbalance (up 1:200), bias, uncertainty in data enhance PAWS, apply our methodology three national parks with diverse characteristics. (i) We use Gaussian processes quantify predictive uncertainty, which exploit improve robustness prescribed patrols increase detection snares average 30%. evaluate on real-world historical from Murchison Falls Queen Elizabeth National Parks Uganda and, first time, Srepok Sanctuary Cambodia. (ii) present results large-scale field tests conducted confirm power PAWS extends promisingly multiple parks. This paper part effort expand 800 around world through integration SMART conservation software.