Bayesian Triplet Loss: Uncertainty Quantification in Image Retrieval
This work is significant for researchers and practitioners in image retrieval who need reliable uncertainty estimates for downstream decision-making, offering an incremental improvement in calibration and efficiency.
This paper addresses the challenge of uncertainty quantification in image retrieval by treating image embeddings as stochastic features. The authors introduce a Bayesian triplet loss, which is a variational approximation of the posterior, achieving state-of-the-art uncertainty estimates and matching the predictive performance of existing methods.
Uncertainty quantification in image retrieval is crucial for downstream decisions, yet it remains a challenging and largely unexplored problem. Current methods for estimating uncertainties are poorly calibrated, computationally expensive, or based on heuristics. We present a new method that views image embeddings as stochastic features rather than deterministic features. Our two main contributions are (1) a likelihood that matches the triplet constraint and that evaluates the probability of an anchor being closer to a positive than a negative; and (2) a prior over the feature space that justifies the conventional l2 normalization. To ensure computational efficiency, we derive a variational approximation of the posterior, called the Bayesian triplet loss, that produces state-of-the-art uncertainty estimates and matches the predictive performance of current state-of-the-art methods.