Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/yaoyao-liu/meta-transfer-learning

TensorFlow and PyTorch implementation of "Meta-Transfer Learning for Few-Shot Learning" (CVPR2019)
https://github.com/yaoyao-liu/meta-transfer-learning

few-shot-learning fewshot-cifar100 meta-learning mini-imagenet tiered-imagenet transfer-learning

Last synced: about 2 months ago
JSON representation

TensorFlow and PyTorch implementation of "Meta-Transfer Learning for Few-Shot Learning" (CVPR2019)

Awesome Lists containing this project

README 

        
# Meta-Transfer Learning for Few-Shot Learning

[![LICENSE](https://img.shields.io/badge/license-MIT-green?style=flat-square)](https://github.com/y2l/meta-transfer-learning-tensorflow/blob/master/LICENSE)

[![Python](https://img.shields.io/badge/python-2.7%20%7C%203.5-blue.svg?style=flat-square&logo=python&color=3776AB)](https://www.python.org/)

[![TensorFlow](https://img.shields.io/badge/tensorflow-1.3.0-orange.svg?style=flat-square&logo=tensorflow&color=FF6F00)](https://github.com/y2l/meta-transfer-learning/tree/master/tensorflow)

[![PyTorch](https://img.shields.io/badge/pytorch-0.4.0-%237732a8?style=flat-square&logo=PyTorch&color=EE4C2C)](https://pytorch.org/)

[![Citations](https://img.shields.io/endpoint?url=https://raw.githubusercontent.com/yaoyao-liu/yaoyao-liu.github.io/google-scholar-stats/gs_data_shieldsio_mtl.json&logo=Google%20Scholar&color=5087ec&style=flat-square&label=citations)](https://scholar.google.com/citations?view_op=view_citation&hl=en&user=Uf9GqRsAAAAJ&citation_for_view=Uf9GqRsAAAAJ:bEWYMUwI8FkC)



This repository contains the TensorFlow and PyTorch implementation for the [CVPR 2019](http://cvpr2019.thecvf.com/) Paper ["Meta-Transfer Learning for Few-Shot Learning"](http://openaccess.thecvf.com/content_CVPR_2019/papers/Sun_Meta-Transfer_Learning_for_Few-Shot_Learning_CVPR_2019_paper.pdf) by [Qianru Sun](https://qianrusun1015.github.io),\* [Yaoyao Liu](https://people.mpi-inf.mpg.de/~yaliu/),\* [Tat-Seng Chua](https://www.chuatatseng.com/), and [Bernt Schiele](https://www.mpi-inf.mpg.de/departments/computer-vision-and-multimodal-computing/people/bernt-schiele/) (\*=equal contribution).



If you have any questions on this repository or the related paper, feel free to [create an issue](https://github.com/yaoyao-liu/meta-transfer-learning/issues/new) or [send me an email](mailto:[email protected]). 



#### Summary



* [Introduction](#introduction)

* [Getting Started](#getting-started)

* [Datasets](#datasets)

* [Performance](#performance)

* [Citation](#citation)

* [Acknowledgements](#acknowledgements)



## Introduction



Meta-learning has been proposed as a framework to address the challenging few-shot learning setting. The key idea is to leverage a large number of similar few-shot tasks in order to learn how to adapt a base-learner to a new task for which only a few labeled samples are available. As deep neural networks (DNNs) tend to overfit using a few samples only, meta-learning typically uses shallow neural networks (SNNs), thus limiting its effectiveness. In this paper we propose a novel few-shot learning method called ***meta-transfer learning (MTL)*** which learns to adapt a ***deep NN*** for ***few shot learning tasks***. Specifically, meta refers to training multiple tasks, and transfer is achieved by learning scaling and shifting functions of DNN weights for each task. We conduct experiments using (5-class, 1-shot) and (5-class, 5-shot) recognition tasks on two challenging few-shot learning benchmarks: π‘šπ‘–π‘›π‘–ImageNet and Fewshot-CIFAR100. 





    




> Figure: Meta-Transfer Learning. (a) Parameter-level fine-tuning (FT) is a conventional meta-training operation, e.g. in MAML. Its update works for all neuron parameters, π‘Š and 𝑏. (b) Our neuron-level scaling and shifting (SS) operations in meta-transfer learning. They reduce the number of learning parameters and avoid overfitting problems. In addition, they keep large-scale trained parameters (in yellow) frozen, preventing β€œcatastrophic forgetting”.



## Getting Started



Please see `README.md` files in the corresponding folders:



* TensorFlow: [\[Document\]](https://github.com/y2l/meta-transfer-learning/blob/master/tensorflow/README.md)

* PyTorch: [\[Document\]](https://github.com/y2l/meta-transfer-learning/blob/master/pytorch/README.md)



## Datasets



Directly download processed images: [\[Download Page\]](https://mtl.yyliu.net/download/)



### π’Žπ’Šπ’π’ŠImageNet



The π‘šπ‘–π‘›π‘–ImageNet dataset was proposed by [Vinyals et al.](http://papers.nips.cc/paper/6385-matching-networks-for-one-shot-learning.pdf) for few-shot learning evaluation. Its complexity is high due to the use of ImageNet images but requires fewer resources and infrastructure than running on the full [ImageNet dataset](https://arxiv.org/pdf/1409.0575.pdf). In total, there are 100 classes with 600 samples of 84Γ—84 color images per class. These 100 classes are divided into 64, 16, and 20 classes respectively for sampling tasks for meta-training, meta-validation, and meta-test. To generate this dataset from ImageNet, you may use the repository [π‘šπ‘–π‘›π‘–ImageNet tools](https://github.com/y2l/mini-imagenet-tools).



### Fewshot-CIFAR100



Fewshot-CIFAR100 (FC100) is based on the popular object classification dataset CIFAR100. The splits were

proposed by [TADAM](https://arxiv.org/pdf/1805.10123.pdf). It offers a more challenging scenario with lower image resolution and more challenging meta-training/test splits that are separated according to object super-classes. It contains 100 object classes and each class has 600 samples of 32 Γ— 32 color images. The 100 classes belong to 20 super-classes. Meta-training data are from 60 classes belonging to 12 super-classes. Meta-validation and meta-test sets contain 20 classes belonging to 4 super-classes, respectively.



### π’•π’Šπ’†π’“π’†π’…ImageNet



The [π‘‘π‘–π‘’π‘Ÿπ‘’π‘‘ImageNet](https://arxiv.org/pdf/1803.00676.pdf) dataset is a larger subset of ILSVRC-12 with 608 classes (779,165 images) grouped into 34 higher-level nodes in the ImageNet human-curated hierarchy. To generate this dataset from ImageNet, you may use the repository π‘‘π‘–π‘’π‘Ÿπ‘’π‘‘ImageNet dataset: [π‘‘π‘–π‘’π‘Ÿπ‘’π‘‘ImageNet tools](https://github.com/y2l/tiered-imagenet-tools). 



## Performance 



|          (%)           | π‘šπ‘–π‘›π‘– 1-shot  | π‘šπ‘–π‘›π‘– 5-shot  | FC100 1-shot | FC100 5-shot | 

| ---------------------- | ------------ | ------------ | ------------ | ------------ |

| `MTL Paper`            | `60.2 Β± 1.8` | `74.3 Β± 0.9` | `43.6 Β± 1.8` | `55.4 Β± 0.9` |

| `TensorFlow`           | `60.8 Β± 1.8` | `74.3 Β± 0.9` | `44.3 Β± 1.8` | `56.8 Β± 1.0` |

* The performance for the PyTorch version is under checking.



## Citation



Please cite our paper if it is helpful to your work:



```bibtex

@inproceedings{SunLCS2019MTL,

  author    = {Qianru Sun and

               Yaoyao Liu and

               Tat{-}Seng Chua and

               Bernt Schiele},

  title     = {Meta-Transfer Learning for Few-Shot Learning},

  booktitle = {{IEEE} Conference on Computer Vision and Pattern Recognition, {CVPR}

               2019, Long Beach, CA, USA, June 16-20, 2019},

  pages     = {403--412},

  publisher = {Computer Vision Foundation / {IEEE}},

  year      = {2019}

}

```



## Acknowledgements



Our implementations use the source code from the following repositories and users:



* [Model-Agnostic Meta-Learning](https://github.com/cbfinn/maml)



* [Optimization as a Model for Few-Shot Learning](https://github.com/gitabcworld/FewShotLearning)



* [Learning Embedding Adaptation for Few-Shot Learning](https://github.com/Sha-Lab/FEAT)



* [dragen1860/MAML-Pytorch](https://github.com/dragen1860/MAML-Pytorch)



* [@icoz69](https://github.com/icoz69)



* [@CookieLau](https://github.com/CookieLau)