---
OA_place: repository
_id: '18224'
abstract:
- lang: eng
text: Learning from one or few visual examples is one of the key capabilities of
humans since early infancy, but is still a significant challenge for modern AI
systems. While considerable progress has been achieved in few-shot learning from
a few image examples, much less attention has been given to the verbal descriptions
that are usually provided to infants when they are presented with a new object.
In this paper, we focus on the role of additional semantics that can significantly
facilitate few-shot visual learning. Building upon recent advances in few-shot
learning with additional semantic information, we demonstrate that further improvements
are possible by combining multiple and richer semantics (category labels, attributes,
and natural language descriptions). Using these ideas, we offer the community
new results on the popular miniImageNet and CUB few-shot benchmarks, comparing
favorably to the previous state-of-the-art results for both visual only and visual
plus semantics-based approaches. We also performed an ablation study investigating
the components and design choices of our approach. Code available on github.com/EliSchwartz/mutiple-semantics.
article_processing_charge: No
article_type: original
arxiv: 1
author:
- first_name: Eli
full_name: Schwartz, Eli
last_name: Schwartz
- first_name: Leonid
full_name: Karlinsky, Leonid
last_name: Karlinsky
- first_name: Rogerio
full_name: Feris, Rogerio
last_name: Feris
- first_name: Raja
full_name: Giryes, Raja
last_name: Giryes
- first_name: Alexander
full_name: Bronstein, Alexander
id: 58f3726e-7cba-11ef-ad8b-e6e8cb3904e6
last_name: Bronstein
orcid: 0000-0001-9699-8730
citation:
ama: Schwartz E, Karlinsky L, Feris R, Giryes R, Bronstein AM. Baby steps towards
few-shot learning with multiple semantics. Pattern Recognition Letters.
2022;160:142-147. doi:10.1016/j.patrec.2022.06.012
apa: Schwartz, E., Karlinsky, L., Feris, R., Giryes, R., & Bronstein, A. M.
(2022). Baby steps towards few-shot learning with multiple semantics. Pattern
Recognition Letters. Elsevier. https://doi.org/10.1016/j.patrec.2022.06.012
chicago: Schwartz, Eli, Leonid Karlinsky, Rogerio Feris, Raja Giryes, and Alex M.
Bronstein. “Baby Steps towards Few-Shot Learning with Multiple Semantics.” Pattern
Recognition Letters. Elsevier, 2022. https://doi.org/10.1016/j.patrec.2022.06.012.
ieee: E. Schwartz, L. Karlinsky, R. Feris, R. Giryes, and A. M. Bronstein, “Baby
steps towards few-shot learning with multiple semantics,” Pattern Recognition
Letters, vol. 160. Elsevier, pp. 142–147, 2022.
ista: Schwartz E, Karlinsky L, Feris R, Giryes R, Bronstein AM. 2022. Baby steps
towards few-shot learning with multiple semantics. Pattern Recognition Letters.
160, 142–147.
mla: Schwartz, Eli, et al. “Baby Steps towards Few-Shot Learning with Multiple Semantics.”
Pattern Recognition Letters, vol. 160, Elsevier, 2022, pp. 142–47, doi:10.1016/j.patrec.2022.06.012.
short: E. Schwartz, L. Karlinsky, R. Feris, R. Giryes, A.M. Bronstein, Pattern Recognition
Letters 160 (2022) 142–147.
date_created: 2024-10-08T12:54:03Z
date_published: 2022-08-01T00:00:00Z
date_updated: 2024-10-14T10:58:20Z
day: '01'
doi: 10.1016/j.patrec.2022.06.012
extern: '1'
external_id:
arxiv:
- '1906.01905'
fulldoi: https://doi.org/10.1016/j.patrec.2022.06.012
intvolume: ' 160'
language:
- iso: eng
main_file_link:
- open_access: '1'
url: https://arxiv.org/abs/1906.01905
month: '08'
oa: 1
oa_version: Preprint
page: 142-147
publication: Pattern Recognition Letters
publication_identifier:
issn:
- 0167-8655
publication_status: published
publisher: Elsevier
quality_controlled: '1'
scopus_import: '1'
status: public
title: Baby steps towards few-shot learning with multiple semantics
type: journal_article
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
volume: 160
year: '2022'
...
---
OA_place: repository
OA_type: green
_id: '18235'
abstract:
- lang: eng
text: 'Recently, great progress has been made in the field of Few-Shot Learning
(FSL). While many different methods have been proposed, one of the key factors
leading to higher FSL performance is surprisingly simple. It is the backbone network
architecture used to embed the images of the few-shot tasks. While first works
on FSL resorted to small architectures with just a few convolution layers, recent
works show that large architectures pre-trained on the training portion of FSL
datasets produce strong features that are more easily transferable to novel few-shot
tasks, thus attaining significant gains to methods using them. Despite these observations,
little to no work has been done towards finding the right backbone for FSL. In
this paper we propose MetAdapt that not only meta-searches for an optimized architecture
for FSL using Network Architecture Search (NAS), but also results in a model that
can adaptively ‘re-wire’ itself predicting the better architecture for a given
novel few-shot task. Using the proposed approach we observe strong results on
two popular few-shot benchmarks: miniImageNet and FC100.'
article_processing_charge: No
article_type: original
arxiv: 1
author:
- first_name: Sivan
full_name: Doveh, Sivan
last_name: Doveh
- first_name: Eli
full_name: Schwartz, Eli
last_name: Schwartz
- first_name: Chao
full_name: Xue, Chao
last_name: Xue
- first_name: Rogerio
full_name: Feris, Rogerio
last_name: Feris
- first_name: Alexander
full_name: Bronstein, Alexander
id: 58f3726e-7cba-11ef-ad8b-e6e8cb3904e6
last_name: Bronstein
orcid: 0000-0001-9699-8730
- first_name: Raja
full_name: Giryes, Raja
last_name: Giryes
- first_name: Leonid
full_name: Karlinsky, Leonid
last_name: Karlinsky
citation:
ama: 'Doveh S, Schwartz E, Xue C, et al. MetAdapt: Meta-learned task-adaptive architecture
for few-shot classification. Pattern Recognition Letters. 2021;149:130-136.
doi:10.1016/j.patrec.2021.05.010'
apa: 'Doveh, S., Schwartz, E., Xue, C., Feris, R., Bronstein, A. M., Giryes, R.,
& Karlinsky, L. (2021). MetAdapt: Meta-learned task-adaptive architecture
for few-shot classification. Pattern Recognition Letters. Elsevier. https://doi.org/10.1016/j.patrec.2021.05.010'
chicago: 'Doveh, Sivan, Eli Schwartz, Chao Xue, Rogerio Feris, Alex M. Bronstein,
Raja Giryes, and Leonid Karlinsky. “MetAdapt: Meta-Learned Task-Adaptive Architecture
for Few-Shot Classification.” Pattern Recognition Letters. Elsevier, 2021.
https://doi.org/10.1016/j.patrec.2021.05.010.'
ieee: 'S. Doveh et al., “MetAdapt: Meta-learned task-adaptive architecture
for few-shot classification,” Pattern Recognition Letters, vol. 149. Elsevier,
pp. 130–136, 2021.'
ista: 'Doveh S, Schwartz E, Xue C, Feris R, Bronstein AM, Giryes R, Karlinsky L.
2021. MetAdapt: Meta-learned task-adaptive architecture for few-shot classification.
Pattern Recognition Letters. 149, 130–136.'
mla: 'Doveh, Sivan, et al. “MetAdapt: Meta-Learned Task-Adaptive Architecture for
Few-Shot Classification.” Pattern Recognition Letters, vol. 149, Elsevier,
2021, pp. 130–36, doi:10.1016/j.patrec.2021.05.010.'
short: S. Doveh, E. Schwartz, C. Xue, R. Feris, A.M. Bronstein, R. Giryes, L. Karlinsky,
Pattern Recognition Letters 149 (2021) 130–136.
date_created: 2024-10-08T12:57:53Z
date_published: 2021-09-01T00:00:00Z
date_updated: 2024-10-15T07:39:56Z
day: '01'
doi: 10.1016/j.patrec.2021.05.010
extern: '1'
external_id:
arxiv:
- '1912.00412'
fulldoi: https://doi.org/10.1016/j.patrec.2021.05.010
intvolume: ' 149'
language:
- iso: eng
main_file_link:
- open_access: '1'
url: https://doi.org/10.48550/arXiv.1912.00412
month: '09'
oa: 1
oa_version: Preprint
page: 130-136
publication: Pattern Recognition Letters
publication_identifier:
issn:
- 0167-8655
publication_status: published
publisher: Elsevier
quality_controlled: '1'
scopus_import: '1'
status: public
title: 'MetAdapt: Meta-learned task-adaptive architecture for few-shot classification'
type: journal_article
user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87
volume: 149
year: '2021'
...