Honors College Thesis
 

Transformers for representing microbiome data: an empirical evaluation

Public Deposited

Downloadable Content

Download PDF
https://ir.library.oregonstate.edu/concern/honors_college_theses/pn89df65n

Descriptions

Attribute NameValues
Creator
Abstract
  • As the link between human microbiomes and health has become more established, the interest in applying statistical approaches to microbiome data to understand the mechanisms behind these links has grown. However, microbiome data is often of unmanageable size, and consequently, producing quality lower dimensional representations of samples is a significant challenge in microbiome analysis. One approach to combatting these challenges is by leveraging the similarities between microbiome data and natural language data, and previous work applying natural language embedding algorithms to the microbiome domain has shown promise. In this work, we explore the effectiveness of applying transformers, the current state of the art in the natural language domain to microbiome data. We find that transformers can produce sample representations that outperform previous approaches. We also demonstrate that pretraining schemes from the natural language domain are effective in teaching transformers transferrable information that improves performance on downstream microbiome tasks.
Contributor
Resource Type
Date Issued
Degree Level
Degree Name
Degree Field
Degree Grantor
Commencement Year
Advisor
Committee Member
Non-Academic Affiliation
Rights Statement
Related Items
  • Project Code
Publisher
Peer Reviewed
Language
File Extent
  • 21 pages
Embargo reason
  • Pending Publication
Embargo date range
  • 2021-08-25 to 2021-11-07

Relationships

Parents:

This work has no parents.

In Collection:

Items

Thumbnail Title Date Uploaded Visibility Actions
file details RohanVarma2021.pdf 2021-08-25 Public Download