In this paper, we study the problem of Novel Class Discovery (NCD). NCD aims at inferring novel object categories in an unlabeled set by leveraging from prior knowledge of a labeled set containing different, but related classes. Existing approaches tackle this problem by considering multiple objective functions, usually involving specialized loss terms for the labeled and the unlabeled samples respectively, and often requiring auxiliary regularization terms. In this paper we depart from this traditional scheme and introduce a UNified Objective function (UNO) for discovering novel classes, with the explicit purpose of favoring synergy between supervised and unsupervised learning. Using a multi-view self-labeling strategy, we generate pseudo-labels that can be treated homogeneously with ground truth labels. This leads to a single classification objective operating on both known and unknown classes. Despite its simplicity, UNO outperforms the state of the art by a significant margin on several benchmarks (≈+10% on CIFAR-100 and +8% on ImageNet). The project page is available at: https://ncd-uno.github.io.
A Unified Objective for Novel Class Discovery / Fini, E.; Sangineto, E.; Lathuiliere, S.; Zhong, Z.; Nabi, M.; Ricci, E.. - (2021), pp. 9264-9272. ((Intervento presentato al convegno 18th IEEE/CVF International Conference on Computer Vision, ICCV 2021 tenutosi a Virtual nel 2021 [10.1109/ICCV48922.2021.00915].
Scheda prodotto non validato
I dati visualizzati non sono stati ancora sottoposti a validazione formale da parte dello Staff di IRIS, ma sono stati ugualmente trasmessi al Sito Docente Cineca (Loginmiur).
Titolo: | A Unified Objective for Novel Class Discovery | |
Autori: | Fini, E.; Sangineto, E.; Lathuiliere, S.; Zhong, Z.; Nabi, M.; Ricci, E. | |
Autori Unitn: | ||
Titolo del volume contenente il saggio: | Proceedings of the IEEE International Conference on Computer Vision | |
Luogo di edizione: | Virtual | |
Casa editrice: | Institute of Electrical and Electronics Engineers Inc. | |
Anno di pubblicazione: | 2021 | |
Codice identificativo Scopus: | 2-s2.0-85127814075 | |
ISBN: | 978-1-6654-2812-5 | |
Handle: | http://hdl.handle.net/11572/341688 | |
Citazione: | A Unified Objective for Novel Class Discovery / Fini, E.; Sangineto, E.; Lathuiliere, S.; Zhong, Z.; Nabi, M.; Ricci, E.. - (2021), pp. 9264-9272. ((Intervento presentato al convegno 18th IEEE/CVF International Conference on Computer Vision, ICCV 2021 tenutosi a Virtual nel 2021 [10.1109/ICCV48922.2021.00915]. | |
Appare nelle tipologie: |