BEGIN:VCALENDAR
VERSION:2.0
PRODID:icalendar-ruby
CALSCALE:GREGORIAN
X-WR-CALNAME:Thesis Defense: Nathan Sundquist
X-WR-TIMEZONE:Mountain Time (US & Canada)
BEGIN:VEVENT
DTSTAMP:20261014T222838Z
UID:tag:localist.com\,2008:EventInstance_53603153449087
DTSTART:20260803T190000Z
DTEND:20260803T200000Z
DESCRIPTION:Title: Topological Analysis Of Wildfire Ignition Data\n\nProgra
 m: Mathematics MS\n\nCommittee Chair: Michael Perlmutter\n\nCommitee: Mich
 ael Perlmutter\, Zach Teitler\, Jens Harlander\n\nAbstract: Topological Da
 ta Analysis (TDA) applies techniques from algebraic topology to extract qu
 alitative geometric and topological features from complex datasets. This t
 hesis presents foundational concepts in TDA\, including simplicial complex
 es\, nerves\, filtrations\, and persistent homology. It then demonstrates 
 their application to problems in public resource allocation and wildfire c
 lassification. Coverage maps showing public accessibility to cooling cente
 rs in Austin\, Texas and Boise\, Idaho are created by performing persisten
 t homology on witness complexes built from the geospatial data of each cit
 y. These maps can be used as a supplement for the Heat Vulnerability Index
 \, or as a low-cost alternative in regions where demographic data is less 
 readily available. The Mapper algorithm is applied to the high-dimensional
  Fire Program Analysis Fire-Occurrence Database-Attributes dataset to prod
 uce graph representations that preserve topological structure while provid
 ing interpretable visualizations of the data. Several machine learning mod
 els are then used for the classification of wildfire ignition causes. Tree
 -based methods\, particularly XGBoost\, achieve higher predictive accuracy
  and lower computational cost than the neural network architectures consid
 ered. These results illustrate the utility of topological methods for both
  exploratory data analysis and practical decision making\, and suggest tha
 t the integration of topological features\, using methods such as persiste
 nce landscapes\, provides a promising direction for the enhancement of mac
 hine learning models in future work.
GEO:43.604437;-116.205673
LOCATION:Mathematics Building (MB)\, 126
SUMMARY:Thesis Defense: Nathan Sundquist
URL;VALUE=URI:https://events.boisestate.edu/event/thesis-defense-nathan-sun
 dquist
CATEGORIES:Lectures and Presentations
END:VEVENT
END:VCALENDAR