Bootstrapping Parallel Anchors for Relative Representations
This work solves the problem of reducing anchor dependency for researchers and practitioners using relative representations, though it is incremental as it builds on existing relative representation frameworks.
The paper addresses the impracticality of requiring many parallel anchors for relative representations by proposing an optimization-based method to discover new anchors from a limited seed set, achieving competitive results in tasks like semantic correspondence and space alignment.
The use of relative representations for latent embeddings has shown potential in enabling latent space communication and zero-shot model stitching across a wide range of applications. Nevertheless, relative representations rely on a certain amount of parallel anchors to be given as input, which can be impractical to obtain in certain scenarios. To overcome this limitation, we propose an optimization-based method to discover new parallel anchors from a limited known set (seed). Our approach can be used to find semantic correspondence between different domains, align their relative spaces, and achieve competitive results in several tasks.