Source-free domain adaptation (SFDA) aims to adapt a classifier to an unlabelled target data set by only using a pre-trained source model. However, the absence of the source data and the domain shift makes the predictions on the target data unreliable. We propose quantifying the uncertainty in the source model predictions and utilizing it to guide the target adaptation. For this, we construct a probabilistic source model by incorporating priors on the network parameters inducing a distribution over the model predictions. Uncertainties are estimated by employing a Laplace approximation and incorporated to identify target data points that do not lie in the source manifold and to down-weight them when maximizing the mutual information on the target data. Unlike recent works, our probabilistic treatment is computationally lightweight, decouples source training and target adaptation, and requires no specialized source training or changes of the model architecture. We show the advantages of uncertainty-guided SFDA over traditional SFDA in the closed-set and open-set settings and provide empirical evidence that our approach is more robust to strong domain shifts even without tuning.

Uncertainty-Guided Source-Free Domain Adaptation / Roy, Subhankar; Trapp, Martin; Pilzer, Andrea; Kannala, Juho; Sebe, Nicu; Ricci, Elisa; Solin, Arno. - 13685:(2022), pp. 537-555. (Intervento presentato al convegno European Conference on Computer Vision (ECCV) tenutosi a Tel Aviv nel 23-27 October 2022) [10.1007/978-3-031-19806-9_31].

Uncertainty-Guided Source-Free Domain Adaptation

Roy, Subhankar;Pilzer, Andrea;Sebe, Nicu;Ricci, Elisa;
2022-01-01

Abstract

Source-free domain adaptation (SFDA) aims to adapt a classifier to an unlabelled target data set by only using a pre-trained source model. However, the absence of the source data and the domain shift makes the predictions on the target data unreliable. We propose quantifying the uncertainty in the source model predictions and utilizing it to guide the target adaptation. For this, we construct a probabilistic source model by incorporating priors on the network parameters inducing a distribution over the model predictions. Uncertainties are estimated by employing a Laplace approximation and incorporated to identify target data points that do not lie in the source manifold and to down-weight them when maximizing the mutual information on the target data. Unlike recent works, our probabilistic treatment is computationally lightweight, decouples source training and target adaptation, and requires no specialized source training or changes of the model architecture. We show the advantages of uncertainty-guided SFDA over traditional SFDA in the closed-set and open-set settings and provide empirical evidence that our approach is more robust to strong domain shifts even without tuning.
2022
European Conference on Computer Vision (ECCV)
Heidelberg
Springer
978-3-031-19805-2
978-3-031-19806-9
Roy, Subhankar; Trapp, Martin; Pilzer, Andrea; Kannala, Juho; Sebe, Nicu; Ricci, Elisa; Solin, Arno
Uncertainty-Guided Source-Free Domain Adaptation / Roy, Subhankar; Trapp, Martin; Pilzer, Andrea; Kannala, Juho; Sebe, Nicu; Ricci, Elisa; Solin, Arno. - 13685:(2022), pp. 537-555. (Intervento presentato al convegno European Conference on Computer Vision (ECCV) tenutosi a Tel Aviv nel 23-27 October 2022) [10.1007/978-3-031-19806-9_31].
File in questo prodotto:
File Dimensione Formato  
136850530 (1).pdf

Open Access dal 21/10/2023

Tipologia: Post-print referato (Refereed author’s manuscript)
Licenza: Tutti i diritti riservati (All rights reserved)
Dimensione 1.13 MB
Formato Adobe PDF
1.13 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/11572/361311
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 5
  • ???jsp.display-item.citation.isi??? 7
social impact