Off-campus UMass Amherst users: To download campus access dissertations, please use the following link to log into our proxy server with your UMass Amherst user name and password.

Non-UMass Amherst users: Please talk to your librarian about requesting this dissertation through interlibrary loan.

Dissertations that have an embargo placed on them will not be available to anyone until the embargo expires.

Author ORCID Identifier

https://orcid.org/0000-0002-8838-8905

Document Type

Open Access Dissertation

Degree Name

Doctor of Philosophy (PhD)

Degree Program

Computer Science

Year Degree Awarded

2020

Month Degree Awarded

February

First Advisor

Daniel Sheldon

Second Advisor

Justin Domke

Third Advisor

Benjamin Marlin

Fourth Advisor

Christopher Sutherland

Subject Categories

Computer Sciences

Abstract

Modeling the interactions of different population dynamics (e.g. reproduction, migration) within a population is a challenging problem that underlies numerous ecological research questions. Powerful, interpretable models for population dynamics are key to developing intervention tactics, allocating limited conservation resources, and predicting the impact of uncertain environmental forces on a population. Fortunately, probabilistic graphical models provide a robust mechanistic framework for these kinds of problems. However, in the relatively common case where individuals in the population are unmarked (i.e. indistinguishable from one another), models of the population dynamics naturally contain a deceptively challenging statistical feature: discrete latent variables with unbounded/countably infinite support. Unfortunately, existing inference algorithms for discrete distributions are applicable only for finite distributions and while approximate inference algorithms exist for countably infinite discrete distributions, they are generally unreliable and inefficient. In this work, we develop the first known general-purpose polynomial-time exact inference algorithms for this class of models using a novel representation based on probability generating functions. These methods are flexibe, easy to use, and significantly faster than existing approximate solutions. We also introduce a novel approximation scheme based on this technique that allows it to gracefully scale to populations well beyond the computational limits of any previously known exact or approximate general-purpose inference algorithm for population dynamics. Finally, we conduct an ecological case study on historical data demonstrating the downstream impact of these advances to a large scale population monitoring setting.

Share

COinS