Orthogonal Alignment Methods for Cross-Domain Data Analysis

This research explores data alignment focusing on its motivation, challenges, and applications. It reviews Procrustes Analysis in shape comparison, Wahba’s problem in satellite orientation, and their connections. By embedding alignment methods in shared spaces, it discusses optimisation for cross-domain tasks like word embeddings. A unified orthogonal framework is proposed, highlighting universal principles for diverse domains, ensuring efficient alignment utilities.