Publication Date
Spring 2024
Degree Type
Master's Project
Degree Name
Master of Science in Computer Science (MSCS)
Department
Computer Science
First Advisor
Fabio Di Troia
Second Advisor
Faranak Abri
Third Advisor
William Andreopoulos
Keywords
Domain Switch, Sentiment Analysis, Gradient Reversal Layer
Abstract
Switching domains in sentiment analysis presents the challenge of transferring learned knowledge from one context to another without the need to label data. Traditional methods often struggle when dealing with differences in data distribution a problem known as the domain shift issue. To tackle this using Gradient Reversal Layers (GRL) has emerged as a solution for adapting to different domains in an unsupervised learning setting. This study introduces an enhancement to the standard GRL approach by incorporating a sigmoid function that gradually adjusts how intensely domain adaptation occurs during training. This upgraded GRL technique ensures controlled learning outcomes making it easier for knowledge to transition smoothly between domains. By utilizing sentiment datasets from Amazon Electronics and IMDb reviews the research confirms that this new method outperforms existing approaches. The findings indicate an enhancement in the adaptability and accuracy of sentiment analysis models across diverse domains providing a competitive edge, for industries where flexibility and rapid implementation are crucial.
Recommended Citation
Veeraboina, Hemish, "Domain Switch on Sentiment Analysis using Gradient Reversal Layer" (2024). Master's Projects. 1384.
DOI: https://doi.org/10.31979/etd.hjed-n7kr
https://scholarworks.sjsu.edu/etd_projects/1384
