A Robust Negative Learning Approach to Partial Domain Adaptation Using Source Prototypes

TL;DR

This paper introduces a robust partial domain adaptation framework that uses ensemble learning, prototype-based distribution alignment, and source data privacy preservation to improve adaptation performance and robustness.

Contribution

It proposes a novel PDA method combining ensemble target supervision, prototype-based distribution alignment, and source privacy preservation, which outperforms existing methods.

Findings

Enhanced robustness and generalization demonstrated on benchmark datasets.

Outperforms state-of-the-art PDA approaches in experiments.

Effective pseudo-label refinement and negative transfer mitigation.

Abstract

This work proposes a robust Partial Domain Adaptation (PDA) framework that mitigates the negative transfer problem by incorporating a robust target-supervision strategy. It leverages ensemble learning and includes diverse, complementary label feedback, alleviating the effect of incorrect feedback and promoting pseudo-label refinement. Rather than relying exclusively on first-order moments for distribution alignment, our approach offers explicit objectives to optimize intra-class compactness and inter-class separation with the inferred source prototypes and highly-confident target samples in a domain-invariant fashion. Notably, we ensure source data privacy by eliminating the need to access the source data during the adaptation phase through a priori inference of source prototypes. We conducted a series of comprehensive experiments, including an ablation analysis, covering a range of partial domain adaptation tasks. Comprehensive evaluations on benchmark datasets corroborate our framework's enhanced robustness and generalization, demonstrating its superiority over existing state-of-the-art PDA approaches.