Enhancing Representation Learning with Deep Classifiers in Presence of Shortcut

Ahmadian, Amirhossein,1992-Linköpings universitet,Statistik och maskininlärning,Tekniska fakulteten (författare)

A deep neural classifier trained on an upstream task can be leveraged to boost the performance of another classifier in a related downstream task through the representations learned in hidden layers. However, presence of shortcuts (easy-to-learn features) in the upstream task can considerably impair the versatility of intermediate representations and, in turn, the downstream performance. In this paper, we propose a method to improve the representations learned by deep neural image classifiers in spite of a shortcut in upstream data. In our method, the upstream classification objective is augmented with a type of adversarial training where an auxiliary network, so called lens, fools the classifier by exploiting the shortcut in reconstructing images. Empirical comparisons in self-supervised and transfer learning problems with three shortcut-biased datasets suggest the advantages of our method in terms of downstream performance and/or training time.

Lindsten, Fredrik,1984-Linköpings universitet,Reglerteknik,Statistik och maskininlärning,Tekniska fakulteten(Swepub:liu)freli29 (författare)
Linköpings universitetStatistik och maskininlärning (creator_code:org_t)

Om ämnet

NATURVETENSKAP: NATURVETENSKAP; och Data och informa ...; och Datorseende och ...

Kungliga biblioteket hanterar dina personuppgifter i enlighet med EU:s dataskyddsförordning (2018), GDPR. Läs mer om hur det funkar här.
Så här hanterar KB dina uppgifter vid användning av denna tjänst.