Self-Supervised Prototypical Transfer Learning for Few-Shot Classification

Published in

International Conference on Machine Learning(2020)

External Links:

Generate Graph

TL;DR

It is demonstrated that the self-supervised prototypical transfer learning approach ProtoTransfer outperforms state-of-the-art unsupervised meta-learning methods on few-shot tasks from the mini-ImageNet dataset and has comparable performance to supervised methods, but requires orders of magnitude fewer labels.

Abstract

Most approaches in few-shot learning rely on costly annotated data related to the goal task domain during (pre-)training. Recently, unsupervised meta-learning methods have exchanged the annotation requirement for a reduction in few-shot classification performance. Simultaneously, in settings with realistic domain shift, common transfer learning has been shown to outperform supervised meta-learning. Building on these insights and on advances in self-supervised learning, we propose a transfer learning approach which constructs a metric embedding that clusters unlabeled prototypical samples and their augmentations closely together. This pre-trained embedding is a starting point for few-shot classification by summarizing class clusters and fine-tuning. We demonstrate that our self-supervised prototypical transfer learning approach ProtoTransfer outperforms state-of-the-art unsupervised meta-learning methods on few-shot tasks from the mini-ImageNet dataset. In few-shot experiments with domain shift, our approach even has comparable performance to supervised methods, but requires orders of magnitude fewer labels.

Authors

Carlos Medina

1 papers

A. Devos

1 papers

M. Grossglauser

1 papers

References46 items

What makes for good views for contrastive learning

Prototypical Contrastive Learning of Unsupervised Representations

Unsupervised Few-shot Learning via Distribution Shift-based Augmentation

A Simple Framework for Contrastive Learning of Visual Representations

A New Benchmark for Evaluation of Cross-Domain Few-Shot Learning

Self-Supervised Prototypical Transfer Learning for Few-Shot Classification

Published in

International Conference on Machine Learning(2020)

External Links:

Generate Graph

TL;DR

Abstract

Authors

Carlos Medina

1 papers

A. Devos

1 papers

M. Grossglauser

1 papers

References46 items

What makes for good views for contrastive learning

Prototypical Contrastive Learning of Unsupervised Representations

Unsupervised Few-shot Learning via Distribution Shift-based Augmentation

A Simple Framework for Contrastive Learning of Visual Representations

A New Benchmark for Evaluation of Cross-Domain Few-Shot Learning

Self-Supervised Learning for Few-Shot Image Classification

Momentum Contrast for Unsupervised Visual Representation Learning

When Does Self-supervision Improve Few-shot Learning?

Unsupervised Few-shot Learning via Self-supervised Training

Boosting Few-Shot Visual Learning With Self-Supervision

AutoAugment: Learning Augmentation Strategies From Data

Subspace Networks for Few-shot Classification

A Closer Look at Few-shot Classification

Unsupervised Embedding Learning via Invariant and Spreading Instance Feature

Meta-Dataset: A Dataset of Datasets for Learning to Learn from Few Examples

Assume, Augment and Learn: Unsupervised Few-Shot Meta-Learning via Random Labels and Data Augmentation

Skin Lesion Analysis Toward Melanoma Detection 2018: A Challenge Hosted by the International Skin Imaging Collaboration (ISIC)

Self-Supervised Generalisation with Meta Auxiliary Learning

Unsupervised Meta-Learning for Few-Shot Image Classification

Unsupervised Learning via Meta-Learning

The HAM10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions

Meta-Learning for Semi-Supervised Few-Shot Classification

Few-Shot Learning with Metric-Agnostic Conditional Embeddings

EuroSAT: A Novel Dataset and Deep Learning Benchmark for Land Use and Land Cover Classification

Random Erasing Data Augmentation

Improved Regularization of Convolutional Neural Networks with Cutout

ChestX-Ray8: Hospital-Scale Chest X-Ray Database and Benchmarks on Weakly-Supervised Classification and Localization of Common Thorax Diseases

Prototypical Networks for Few-shot Learning

Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks

Optimization as a Model for Few-Shot Learning

Matching Networks for One Shot Learning

One-shot Learning with Memory-Augmented Neural Networks

Using Deep Learning for Image-Based Plant Disease Detection

Adam: A Method for Stochastic Optimization

ImageNet Large Scale Visual Recognition Challenge

ImageNet classification with deep convolutional neural networks

Caltech-UCSD Birds 200

One-shot learning of object categories

Random erasing of a rectangular region in an image

Accuracy (%) of methods on N -way K-shot classification tasks of the CDFSL benchmark (Guo et al., 2019). All models are trained on mini-ImageNet with ResNet-10. Results style: best and second best

Realistic Evaluation of Deep Semi-Supervised Learning Algorithms

Omniglot transforms For Omniglot we use a set of custom augmentations, namely random resizing and cropping, horizontal and vertical flipping, Image-Pixel Dropout

One shot learning of simple visual concepts

Visualizing Data using t-SNE

* Random vertical flip

Each of the t-SNE plots in Figure 3 shows 500 randomly selected embedded images from within those classes

Field of Study

Computer ScienceMathematics

Journal Information

Name

ArXiv

Volume

abs/2005.00687

Venue Information

Name

International Conference on Machine Learning

Type

conference

URL

https://icml.cc/

Alternate Names

ICML
Int Conf Mach Learn