High-quality calibrated uncertainty estimates are crucial for numerous real-world applications, especially for deep learning-based deployed ML systems. While Bayesian deep learning techniques allow uncertainty estimation, training them with large-scale datasets is an expensive process that does not always yield models competitive with non-Bayesian counterparts. Moreover, many of the high-performing deep learning models that are already trained and deployed are non-Bayesian in nature and do not provide uncertainty estimates. To address these issues, we propose BayesCap that learns a Bayesian identity mapping for the frozen model, allowing uncertainty estimation. BayesCap is a memory-efficient method that can be trained on a small fraction of the original dataset, enhancing pretrained non-Bayesian computer vision models by providing calibrated uncertainty estimates for the predictions without (i) hampering the performance of the model and (ii) the need for expensive retraining the model from scratch. The proposed method is agnostic to various architectures and tasks. We show the efficacy of our method on a wide variety of tasks with a diverse set of architectures, including image super-resolution, deblurring, inpainting, and crucial application such as medical image translation. Moreover, we apply the derived uncertainty estimates to detect out-of-distribution samples in critical scenarios like depth estimation in autonomous driving. Code is available at https://github.com/ExplainableML/BayesCap.

BayesCap: Bayesian Identity Cap for Calibrated Uncertainty in Frozen Neural Networks / Upadhyay, U.; Karthik, S.; Chen, Y.; Mancini, M.; Akata, Z.. - 13672:(2022), pp. 299-317. (Intervento presentato al convegno 17th European Conference on Computer Vision, ECCV 2022 tenutosi a Tel Aviv, Israel nel 23–27 October 2022) [10.1007/978-3-031-19775-8_18].

BayesCap: Bayesian Identity Cap for Calibrated Uncertainty in Frozen Neural Networks

Chen, Y.;Mancini, M.;
2022-01-01

Abstract

High-quality calibrated uncertainty estimates are crucial for numerous real-world applications, especially for deep learning-based deployed ML systems. While Bayesian deep learning techniques allow uncertainty estimation, training them with large-scale datasets is an expensive process that does not always yield models competitive with non-Bayesian counterparts. Moreover, many of the high-performing deep learning models that are already trained and deployed are non-Bayesian in nature and do not provide uncertainty estimates. To address these issues, we propose BayesCap that learns a Bayesian identity mapping for the frozen model, allowing uncertainty estimation. BayesCap is a memory-efficient method that can be trained on a small fraction of the original dataset, enhancing pretrained non-Bayesian computer vision models by providing calibrated uncertainty estimates for the predictions without (i) hampering the performance of the model and (ii) the need for expensive retraining the model from scratch. The proposed method is agnostic to various architectures and tasks. We show the efficacy of our method on a wide variety of tasks with a diverse set of architectures, including image super-resolution, deblurring, inpainting, and crucial application such as medical image translation. Moreover, we apply the derived uncertainty estimates to detect out-of-distribution samples in critical scenarios like depth estimation in autonomous driving. Code is available at https://github.com/ExplainableML/BayesCap.
2022
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
GEWERBESTRASSE 11, CHAM, CH-6330, SWITZERLAND
Springer Science and Business Media Deutschland GmbH
978-3-031-19774-1
978-3-031-19775-8
Upadhyay, U.; Karthik, S.; Chen, Y.; Mancini, M.; Akata, Z.
BayesCap: Bayesian Identity Cap for Calibrated Uncertainty in Frozen Neural Networks / Upadhyay, U.; Karthik, S.; Chen, Y.; Mancini, M.; Akata, Z.. - 13672:(2022), pp. 299-317. (Intervento presentato al convegno 17th European Conference on Computer Vision, ECCV 2022 tenutosi a Tel Aviv, Israel nel 23–27 October 2022) [10.1007/978-3-031-19775-8_18].
File in questo prodotto:
File Dimensione Formato  
136720295.pdf

accesso aperto

Tipologia: Post-print referato (Refereed author’s manuscript)
Licenza: Tutti i diritti riservati (All rights reserved)
Dimensione 2.45 MB
Formato Adobe PDF
2.45 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11572/400370
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 5
  • ???jsp.display-item.citation.isi??? 2
social impact