Probabilistic Embeddings Revisited
This work provides insights for researchers in metric learning and verification tasks, though it appears incremental in scope.
The paper analyzed existing probabilistic embedding methods for verification and retrieval tasks, proposing a simple extension that achieved new state-of-the-art results among probabilistic methods, and established a new benchmark for confidence prediction research.
In recent years, deep metric learning and its probabilistic extensions claimed state-of-the-art results in the face verification task. Despite improvements in face verification, probabilistic methods received little attention in the research community and practical applications. In this paper, we, for the first time, perform an in-depth analysis of known probabilistic methods in verification and retrieval tasks. We study different design choices and propose a simple extension, achieving new state-of-the-art results among probabilistic methods. Finally, we study confidence prediction and show that it correlates with data quality, but contains little information about prediction error probability. We thus provide a new confidence evaluation benchmark and establish a baseline for future confidence prediction research. PyTorch implementation is publicly released.