Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

Self-supervised learning on gene expression data

Created by
  • Haebom

Author

Kevin Dradjat, Massinissa Hamidi, Pierre Bartet, Blaise Hanczar

Outline

This paper applies cutting-edge self-supervised learning techniques to reduce reliance on labeled data for predicting phenotypes from gene expression data. We selected three different self-supervised learning methods and applied them to several public gene expression datasets, evaluating their phenotype prediction accuracy. These methods outperform conventional supervised learning models and demonstrate the advantage of reducing reliance on labeled data. We analyze the strengths and limitations of each method and offer recommendations for case-specific applications and future research directions. This is the first study to combine bulk RNA-Seq data with self-supervised learning.

Takeaways, Limitations

•
Takeaways:
◦
Suggesting the possibility of improving the accuracy of phenotype prediction based on gene expression data through self-supervised learning.
◦
Presenting an effective approach to addressing the lack of labeled data.
◦
We validate the applicability of self-supervised learning to the field of bulk RNA-Seq data analysis.
◦
Comparative analysis of the performance of various self-supervised learning methods and recommendations for their application.
•
Limitations:
◦
This analysis is limited to a comparative performance analysis of a specific self-supervised learning method. Further performance comparisons of other self-supervised learning methods are needed.
◦
Lack of in-depth analysis of performance differences depending on the type and characteristics of the dataset used.
◦
Further research is needed on the interpretability and generalization performance of self-supervised learning models.
👍
Made with Slashpage