ICML 2011 Tutorial: Machine Learning in Ecological Science and Environmental Policy

Topic Overview

This tutorial will review opportunities for machine learning research in ecological science and environmental policy. We will present examples of existing and emerging applications of machine learning and challenges for machine learning research.

Target Audience

Machine learning researchers, especially graduate students and junior faculty. We will assume familiarity with graphical models (primarily directed models) and plate notation.

Presenters

Tutorial Slides

Download the tutorial slides

Outline of the Tutorial

  1. Introduction: from data to models to policies

  2. Data acquisition: Sensors, Human Observers, Re-purposing existing data networks.

  3. Ecological Models and Methods for Learning Them

    1. Species distribution models (SDMs)
      • Case study 1: SDMs from Presence-only data using MaxEnt
      • Case study 2: Building SDMs from eBird data: STEM
      • Case study 3: Building SDMs from eBird data: Detection, Expertise
      • Open Problems
      • Inappropriate extrapolation: SDMs and Climate Change

    2. Dynamical Models
      • Case study 1: Collective graphical models of migration
      • Case study 2: Meta-population Models: SPOMSIM
      • Connection to network cascades

  4. Policy Creation and Optimization