Deep Least Squares Alignment for Unsupervised Domain Adaptation
This work addresses domain adaptation for machine learning applications where labeled data is scarce, but it appears incremental as it builds on existing alignment methods with statistical improvements.
The paper tackled the problem of aligning cross-domain distributions in unsupervised domain adaptation by proposing deep least squares alignment (DLSA), which estimates distributions in a latent space and uses adaptation losses to reduce domain discrepancy, resulting in outperforming state-of-the-art methods in experiments.
Unsupervised domain adaptation leverages rich information from a labeled source domain to model an unlabeled target domain. Existing methods attempt to align the cross-domain distributions. However, the statistical representations of the alignment of the two domains are not well addressed. In this paper, we propose deep least squares alignment (DLSA) to estimate the distribution of the two domains in a latent space by parameterizing a linear model. We further develop marginal and conditional adaptation loss to reduce the domain discrepancy by minimizing the angle between fitting lines and intercept differences and further learning domain invariant features. Extensive experiments demonstrate that the proposed DLSA model is effective in aligning domain distributions and outperforms state-of-the-art methods.