Adversarial Support Alignment
This work addresses domain adaptation challenges with label distribution shifts, offering a novel approach for support alignment, though it is incremental as it builds on existing adversarial alignment methods.
The paper tackles the problem of aligning the supports of distributions without matching densities, proposing a method that minimizes a symmetrized relaxed optimal transport cost in a discriminator's 1D space via an adversarial process. Experiments show it is more robust against label distribution shifts in domain adaptation tasks compared to other alignment-based baselines.
We study the problem of aligning the supports of distributions. Compared to the existing work on distribution alignment, support alignment does not require the densities to be matched. We propose symmetric support difference as a divergence measure to quantify the mismatch between supports. We show that select discriminators (e.g. discriminator trained for Jensen-Shannon divergence) are able to map support differences as support differences in their one-dimensional output space. Following this result, our method aligns supports by minimizing a symmetrized relaxed optimal transport cost in the discriminator 1D space via an adversarial process. Furthermore, we show that our approach can be viewed as a limit of existing notions of alignment by increasing transportation assignment tolerance. We quantitatively evaluate the method across domain adaptation tasks with shifts in label distributions. Our experiments show that the proposed method is more robust against these shifts than other alignment-based baselines.