OOS 44 - Machine Enabled Human Learning: Mining Theory from Data

Organizer(s)

• JW

  James Watson

  Associate Professor
  Oregon State University
  Corvallis, Oregon, United States

Co-Organizer(s)

• JA

  Jorge Arroyo Esquivel

  PhD Candidate
  University of California, Davis
  Davis, CA, United States

If machine learning were around in the 1800s, would Verhulst have discovered his model of logistic growth, or if neural nets were in use would the Lotka-Volterra equations have been created to describe predator-prey interactions? Or, would scientists have relied on black box regression models to make their predictions, and missed the simple rules underlying nature? Recent advances in computer science have made it possible for machines to learn the processes underpinning the complex dynamics that we observe in ecosystems. The “automated discovery” of the laws of nature through machine learning is an exciting and new area of growth in ecology, but there are still many lessons to be learned. In this session we will learn about methods such as symbolic regression, Neural Ordinary Differential Equations and Sparse Identification of Nonlinear Dynamics (SINDy), among others. The goal of this session is to inform ecologists of these cutting-edge approaches that leverage both machine learning and ecological theory to improve our understanding of the natural world. Presentations will include descriptions of research that uses Machine Learning to advance new ecology theory, as well as research that uses ecological theory in concert with machine learning to aid in prediction.

Presentations:

• 1:30 PM – 1:45 PM PDT

  Symbolic regression reveals weak stability and stochastic perturbations maintain persistently cyclical kelp-forest dynamics
  

  Presenting Author: Cheyenne Jarman – Oregon State University

  Co-author: Mark Novak – Oregon State University
• 1:45 PM – 2:00 PM PDT

  Inferring Biological Network Dynamics from Data
  

  Presenting Author: Tracey Oellerich – George Mason University

  Co-author: Maria Emelianenko – George Mason University

  Co-author: Mariaelena Pierobon – George Mason University
• 2:00 PM – 2:15 PM PDT

  Translucent Machine Learning for Ecology: Linking Empirical Dynamic Modeling and Recurrent Neural Networks
  

  Presenting Author: Stephan B. Munch, Southwest Fisheries Science Center – NOAA Fisheries
• 2:15 PM – 2:30 PM PDT

  Revealing evolutionary dynamics with neural differential equations
  

  Presenting Author: Willem Bonnaffé – University of Oxford

  Co-author: Tim Coulson, Pr. – University of Oxford
• 2:30 PM – 2:45 PM PDT

  Predicting ecological regime shifts - and post-transition dynamics - from time series
  

  Presenting Author: Darian Sorenson – University of California Davis

  Co-author: Bethany J. Johnson – University of California Santa Cruz

  Co-author: Stephan B. Munch, Southwest Fisheries Science Center – NOAA Fisheries
• 2:45 PM – 3:00 PM PDT

  Autocomplete for Mechanistic Models via Scientific Machine Learning
  

  Presenting Author: Chris Rackauckas – MIT and JuliaHub