Paper Key : IRJ************694
Author: Dr Pankaj Malik,Anirudh Agrawal,Arpita Laad,Ashutosh Saraswat,Astha Maheshwari,Jyonit Singh
Date Published: 08 Jul 2024
Abstract
Multi-Source Domain Adaptation (MSDA) has emerged as a crucial area in machine learning, enabling the transfer of knowledge from multiple source domains to a target domain. However, the presence of noisy or low-quality data in the source domains can significantly hinder the performance of adapted models, leading to poor generalization and robustness. This paper addresses this challenge by exploring various techniques to identify and mitigate the impact of such data. We provide a comprehensive review of existing methods for detecting and handling noisy data in MSDA, including statistical methods, machine learning approaches, and domain-specific techniques. Furthermore, we propose novel strategies such as dynamic noise identification and a robust multi-source adaptation framework that integrates noise mitigation into the learning process. Our methods are evaluated through extensive experiments on multiple benchmark datasets, demonstrating significant improvements in model performance and robustness. The findings of this study offer valuable insights and practical solutions for enhancing the reliability of MSDA in real-world applications.
DOI Requested