Uncertainty in model-agnostic meta-learning using variational inference

Cuong Nguyen, Thanh-Toan Do, Gustavo Carneiro

Research output: Chapter in Book/Report/Conference proceedingConference PaperResearchpeer-review

2 Citations (Scopus)

Abstract

We introduce a new, rigorously-formulated Bayesian meta-learning algorithm that learns a probability distribution of model parameter prior for few-shot learning. The proposed algorithm employs a gradient-based variational inference to infer the posterior of model parameters for a new task. Our algorithm can be applied to any model architecture and can be implemented in various machine learning paradigms, including regression and classification. We show that the models trained with our proposed meta-learning algorithm are well calibrated and accurate, with state-of-the-art calibration and classification results on three few-shot classification benchmarks (Om- niglot, mini-ImageNet and tiered-ImageNet), and competitive results in a multi-modal task-distribution regression.

Original languageEnglish
Title of host publicationProceedings - 2020 IEEE Winter Conference on Applications of Computer Vision (WACV)
EditorsGang Hua, Ming-Yu Liu, Vishal Patel, Walter Scheirer, Ryan Farrell
Place of PublicationPiscataway NJ USA
PublisherIEEE, Institute of Electrical and Electronics Engineers
Pages3079-3089
Number of pages11
ISBN (Electronic)9781728165530, 9781728165523
ISBN (Print)9781728165547
DOIs
Publication statusPublished - 2020
Externally publishedYes
EventIEEE Winter Conference on Applications of Computer Vision 2020 - Snowmass Village, United States of America
Duration: 1 Mar 20205 Mar 2020
https://ieeexplore.ieee.org/xpl/conhome/9087828/proceeding (Proceedings)
https://wacv20.wacv.net (Website)

Publication series

NameProceedings - 2020 IEEE Winter Conference on Applications of Computer Vision, WACV 2020
PublisherThe Institute of Electrical and Electronics Engineers, Inc.
ISSN (Print)2472-6737
ISSN (Electronic)2642-9381

Conference

ConferenceIEEE Winter Conference on Applications of Computer Vision 2020
Abbreviated titleWACV 2020
CountryUnited States of America
CitySnowmass Village
Period1/03/205/03/20
Internet address

Cite this