Computing Reviews
Today's Issue Hot Topics Search Browse Recommended My Account Log In
Review Help
Search
Artificial intelligence and conservation
Fang F., Tambe M., Dilkina B., Plumptre A., Cambridge University Press, New York, NY, 2019. 246 pp. Type: Book (978-1-108464-73-4)
Date Reviewed: Jun 25 2020

This edited volume is the second book in the publisher’s “Artificial Intelligence for Social Good” series. It is a collection of papers on “research advances in AI that benefit the conservation of wildlife, forests, coral reefs, rivers, and other natural resources.”

The introductory chapter, contributed by the editors, summarizes the state of the art in artificial intelligence (AI) and conservation. They present very useful AI techniques for pattern recognition, stochastic analysis, rule-based decision making, artificial neural networks (ANNs), Markov decision processes (MDPs), and Monte Carlo simulations. The rest of the book is organized in two parts: the first part (four chapters) deals with techniques in use, while the second part (seven chapters) presents specific case studies.

In the second chapter, Plumptre covers issues, challenges, and techniques related to wildlife conservation, including the difficulties with enforcing related laws and how an AI-based model can be used. Uganda’s use of an AI-based system, called SMART, illustrates this approach. In chapter 3, “Wildlife Poaching Forecasting Based on Ranger–Collected Data and Evaluation Through Field Tests,” Gholami et al. present INTERCEPT, a framework that models poaching behavior using data common to the work domain to select the work model, subsequently refined by spatiotemporal data to predict poaching episodes. Using INTERCEPT, along with five months of patrolling data from Uganda’s Queen Elizabeth Protected Area (QEPA), the authors developed a decision algorithm called BoostIT that underwent one-month and eight-month trials with encouraging results.

In the next chapter, “Optimal Patrol Planning Against Black-Box Attackers,” Xu et al. present the optimal patrol planning with enhanced randomness (OPERA) algorithm. OPERA generates patrol maps against unknown random (black-box) attacks. The authors evaluate OPERA against QEPA data to show that high numbers of attacks are predictable. Chapter 5, “Automatic Detection of Poachers and Wildlife with UAVs,” by Bondi et al., shows how automated continuous video streams from unmanned aerial vehicles (UAVs) identify poaching events using an AI-based system, systematic poacher detector (SPOT), despite difficulties such as a) parts of UAV flights resulting in different image sizes, b) UAV stabilization resulting in image streams being opaque in parts, and c) low image resolutions due to thermal sensors/camera. SPOT is trained offline using archival data and preset uses, with fast convolutional neural networks (CNNs) for detection and a JavaScript-based user interface called VIOLA for marking and evaluation.

Part 2 begins with “Protecting Coral Reef Ecosystems via Efficient Patrol,” which focuses on the Great Barrier Reef (Australia) marine protected area (MPA). Here, Yin and An propose a patrolling model and suggest a scalable algorithm (CDOG) based on linear programming. In chapter 7, “Simultaneous Optimization of Strategic and Tactical Planning for Environmental Sustainability and Security,” McCarthy et al. present a study on the threatened hardwood forests of Madagascar that would require training and optimizing the efforts of a number of teams. They propose a hierarchical layered algorithm, FORTIFY, based on integer programming and compact graphs (of spatial-temporal observation and action nodes of workspace).

Next, in “Enforcing Environmental Compliance Through Strategically Randomized Factory Inspections,” Ford et al. describe how the NECTAR game-theoretic application can be used in the conservation of Ganga’s water quality, which is affected by a network of tanneries in Kanpur, India. NECTAR is based on a linear programming problem with an MDP model, defined around defender-attacker status state graphs, representing environmental attacks and defense to compute an optimal inspection strategy.

In chapter 9, “Connecting Conservation Research and Implementation: Building a Wildfire Assistant,” McGregor et al. discuss the development of a wildland fire decision support system (WFDSS), including its requirements and constraints. WFDSS is based on a user-defined MDP and attempts to optimize policy using a fire simulation, visualization, black-box optimization, and a model-free Monte Carlo with independencies (MFMCi) model. It recommends fire and action trajectories, noting that despite short-term threats and losses, fires bring huge benefits to the forest and associated ecology. WFDSS requires experimental verification, particularly for visualization fidelity, and policy optimization.

In the next chapter, “Probabilistic Inference with Generating Functions for Animal Population,” Sheldon et al. state that the embedded “default” count in toolkits for animal population estimation is problematic. They summarize recent AI advances that would help. In chapter 11, Xue and Gomes discuss the benefits and issues with “Engaging Citizen Scientists in Data Collection for Conservation.” They describe the eBird application, the Avicaching game, and Bird-Watcher Assistant, a data-logger built around concepts from game theory, machine learning, and combinatorial optimization.

In the last chapter, “Simulator-Defined Markov Decision Processes,” Albers et al. present a case study on managing bioinvasions by nonnative invasive species (IS). The authors “developed a bio-economic MDP for the tamarisk IS problem” with four components: state space; action space; a dynamical model that handles mortality, seed production, seed dispersal, and seed establishment; and a cost model. They point out that more work is required due to inadequate results, despite using a large number of Monte Carlo simulations.

The breadth and depth of the included work is quite impressive. Each chapter is well referenced. This book will appeal to area experts and researchers.

Reviewer:  Anoop Malaviya Review #: CR147003 (2012-0281)
Bookmark and Share
 
General (I.2.0 )
 
 
Earth And Atmospheric Sciences (J.2 ... )
 
 
Physical Sciences And Engineering (J.2 )
 
Would you recommend this review?
yes
no
Other reviews under "General": Date
Artificial experts: social knowledge and intelligent machines
Collins H., MIT Press, Cambridge, MA, 1990. Type: Book (9780262031684)
Apr 1 1991
Catalogue of artificial intelligence techniques
Bundy A., Springer-Verlag New York, Inc., New York, NY, 1990. Type: Book (9780387529592)
Aug 1 1991
Knowledge and inference
Nagao M., Academic Press Prof., Inc., San Diego, CA, 1990. Type: Book (9780125136624)
Oct 1 1991
more...

E-Mail This Printer-Friendly
Send Your Comments
Contact Us
Reproduction in whole or in part without permission is prohibited.   Copyright 1999-2024 ThinkLoud®
Terms of Use
| Privacy Policy