Abstract

Bat call classification is widely used in bat population monitoring in the field of ecology. Since bat populations are susceptible to changes in their surroundings, it is essential to monitor bat populations for purposes of bat protection and bio-environment protection. The purpose of this thesis is to compare the performance of several classification methods applied to a data set extracted from audio recordings for different species of bats in Mexico. The methods under comparison are (i) a nonparametric Bayesian approach using a multinomial probit model with Gaussian process prior; (ii) support vector machines (SVM); (iii) naive Bayes; and (iv) Bayesian additive regression trees (BART). We find that BART achieves the lowest classification error rate.

Committee Chair

Todd Kuffner

Committee Members

Nan Lin, José E. Figueroa-López

Comments

Permanent URL: https://doi.org/10.7936/K7V1247W

Degree

Master of Arts (AM/MA)

Author's Department

Statistics

Author's School

Graduate School of Arts and Sciences

Document Type

Thesis

Date of Award

Spring 5-18-2018

Language

English (en)

Download

Share

COinS