A Neural Network Based Method with Transfer Learning for Genetic Data Analysis

By J. Lin et al

Published on Aug. 10, 2015

Read the original document by opening this link in a new tab.

Table of Contents

1. Introduction
2. Method
3. Real Data Application
3.1 Real Data Application I
3.2 Real Data Application II

Summary

This paper presents a method that combines transfer learning technique with a neural network based method (expectile neural networks) for genetic data analysis. Transfer learning has emerged as a powerful technique in various applications but has been largely ignored in genetic data analysis. The authors demonstrate the performance improvement by applying transfer learning to neural networks, specifically for genetic sequencing data analysis. The paper discusses expectile regression, neural networks, and the integration of these concepts. Real data applications on alcohol and tobacco use-related phenotypes and Alzheimer's disease data show the effectiveness of the proposed method. Results indicate that expectile neural networks with transfer learning outperform models without transfer learning, providing valuable insights for genetic data analysis.