Renewing Iterative Self-labeling Domain Adaptation with Application to the Spine Motion Prediction

TL;DR

This paper introduces Re-ISDA, a novel transfer learning algorithm that enhances domain adaptation by iteratively renewing self-labeling, with applications demonstrated in spine motion prediction.

Contribution

The paper presents a new transfer learning method, Re-ISDA, which improves domain adaptation through iterative self-labeling renewal, addressing distribution differences between training and testing data.

Findings

Re-ISDA outperforms existing domain adaptation methods.

Effective in spine motion prediction tasks.

Demonstrates robustness across different domain shifts.

Abstract

The area of transfer learning comprises supervised machine learning methods that cope with the issue when the training and testing data have different input feature spaces or distributions. In this work, we propose a novel transfer learning algorithm called Renewing Iterative Self-labeling Domain Adaptation (Re-ISDA). In this work, we propose a novel transfer learning algorithm called Renewing Iterative Self-labeling Domain Adaptation (Re-ISDA).