{ "id": "2302.01332", "version": "v1", "published": "2023-02-02T18:59:23.000Z", "updated": "2023-02-02T18:59:23.000Z", "title": "Bayesian Metric Learning for Uncertainty Quantification in Image Retrieval", "authors": [ "Frederik Warburg", "Marco Miani", "Silas Brack", "Soren Hauberg" ], "comment": "Code: https://github.com/FrederikWarburg/bayesian-metric-learning", "categories": [ "cs.LG", "cs.CV" ], "abstract": "We propose the first Bayesian encoder for metric learning. Rather than relying on neural amortization as done in prior works, we learn a distribution over the network weights with the Laplace Approximation. We actualize this by first proving that the contrastive loss is a valid log-posterior. We then propose three methods that ensure a positive definite Hessian. Lastly, we present a novel decomposition of the Generalized Gauss-Newton approximation. Empirically, we show that our Laplacian Metric Learner (LAM) estimates well-calibrated uncertainties, reliably detects out-of-distribution examples, and yields state-of-the-art predictive performance.", "revisions": [ { "version": "v1", "updated": "2023-02-02T18:59:23.000Z" } ], "analyses": { "keywords": [ "bayesian metric learning", "uncertainty quantification", "image retrieval", "first bayesian encoder", "laplacian metric learner" ], "tags": [ "github project" ], "note": { "typesetting": "TeX", "pages": 0, "language": "en", "license": "arXiv", "status": "editable" } } }