We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Leveraging the Feature Distribution in Transfer-based Few-Shot Learning

Abstract: Few-shot classification is a challenging problem due to the uncertainty caused by using few labelled samples. In the past few years, many methods have been proposed to solve few-shot classification, among which transfer-based methods have proved to achieve the best performance. Following this vein, in this paper we propose a novel transfer-based method that builds on two successive steps: 1) preprocessing the feature vectors so that they become closer to Gaussian-like distributions, and 2) leveraging this preprocessing using an optimal-transport inspired algorithm (in the case of transductive settings). Using standardized vision benchmarks, we prove the ability of the proposed methodology to achieve state-of-the-art accuracy with various datasets, backbone architectures and few-shot settings.
Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Cite as: arXiv:2006.03806 [cs.LG]
  (or arXiv:2006.03806v3 [cs.LG] for this version)

Submission history

From: Yuqing Hu [view email]
[v1] Sat, 6 Jun 2020 07:32:08 GMT (36kb,D)
[v2] Thu, 13 Aug 2020 12:43:51 GMT (36kb,D)
[v3] Tue, 26 Jan 2021 10:03:34 GMT (75kb,D)

Link back to: arXiv, form interface, contact.