The code repository for "How to Train Your MAML to Excel in Few-Shot Classification" (Accepted by ICLR 2022) in PyTorch.
If you use any content of this repo for your work, please cite the following bib entry:
@inproceedings{ye2021UNICORN,
author = {Han-Jia Ye and
Wei-Lun Chao},
title = {How to Train Your {MAML} to Excel in Few-Shot Classification},
booktitle = {10th International Conference on Learning Representations ({ICLR})},
year = {2021}
}
Model-agnostic meta-learning (MAML) is arguably the most popular meta-learning algorithm nowadays, given its flexibility to incorporate various model architectures and to be applied to different problems. Nevertheless, its performance on few-shot classification is far behind many recent algorithms dedicated to the problem. In this paper, we point out several key facets of how to train MAML to excel in few-shot classification. First, we find that a large number of gradient steps are needed for the inner loop update, which contradicts the common usage of MAML for few-shot classification. Second, we find that MAML is sensitive to the permutation of class assignments in meta-testing: for a few-shot task of N classes, there are exponentially many ways to assign the learned initialization of the N-way classifier to the N classes, leading to an unavoidably huge variance. Third, we investigate several ways for permutation invariance and find that learning a shared classifier initialization for all the classes performs the best. On benchmark datasets such as MiniImageNet and TieredImageNet, our approach, which we name UNICORN-MAML, performs on a par with or even outperforms state-of-the-art algorithms, while keeping the simplicity of MAML without adding any extra sub-networks.
Experimental results on few-shot learning datasets with ResNet-12 backbone (Same as the MetaOptNet). We report average results with 10,000 randomly sampled few-shot learning episodes for stablized evaluation.
MiniImageNet Dataset
Setups | 1-Shot 5-Way | 5-Shot 5-Way |
---|---|---|
ProtoMAML | 62.62 | 79.24 |
MetaOptNet | 62.64 | 78.63 |
DeepEMD | 65.91 | 82.41 |
FEAT | 66.78 | 82.05 |
MAML | 64.42 | 83.44 |
UNICORN-MAML | 65.17 | 84.30 |
TieredImageNet Dataset
Setups | 1-Shot 5-Way | 5-Shot 5-Way |
---|---|---|
ProtoMAML | 67.10 | 81.18 |
MetaOptNet | 65.99 | 81.56 |
DeepEMD | 71.52 | 86.03 |
FEAT | 70.80 | 84.79 |
MAML | 65.72 | 84.37 |
UNICORN-MAML | 69.24 | 86.06 |
The following packages are required to run the scripts:
-
Package tensorboardX
-
Dataset: please download the dataset and put images into the folder data/[name of the dataset, miniimagenet or cub]/images
-
Pre-trained weights: The pre-trained weights (used for initialization) could be downloaded at here.
The MiniImageNet dataset is a subset of the ImageNet that includes a total number of 100 classes and 600 examples per class. We follow the previous setup, and use 64 classes as base categories, 16 and 20 as two sets of novel categories for model validation and evaluation, respectively.
TieredImageNet is a large-scale dataset with more categories, which contains 351, 97, and 160 categoriesfor model training, validation, and evaluation, respectively.
To reproduce our experiments with UNICORN-MAML, please use train_fsl.py. There are four parts in the code.
model
: It contains the main files of the code, including the few-shot learning trainer, the dataloader, the network architectures, and baseline and comparison models.data
: Images and splits for the data sets.saves
: The pre-trained weights of different networks.checkpoints
: To save the trained models.
Please use train_fsl.py and follow the instructions below. The file will automatically evaluate the model on the meta-test set with 10,000 tasks after given epochs.
The train_fsl.py takes the following command line options (details are in the model/utils.py
):
Task Related Arguments
-
dataset
: Option for the dataset (MiniImageNet
,TieredImageNet
, orCUB
), default toMiniImageNet
-
way
: The number of classes in a few-shot task during meta-training, default to5
-
eval_way
: The number of classes in a few-shot task during meta-test, default to5
-
shot
: Number of instances in each class in a few-shot task during meta-training, default to1
-
eval_shot
: Number of instances in each class in a few-shot task during meta-test, default to1
-
query
: Number of instances in each class to evaluate the performance during meta-training, default to15
-
eval_query
: Number of instances in each class to evaluate the performance during meta-test, default to15
Optimization Related Arguments
-
max_epoch
: The maximum number of training epochs, default to200
-
episodes_per_epoch
: The number of tasks sampled in each epoch, default to100
-
num_eval_episodes
: The number of tasks sampled from the meta-val set to evaluate the performance of the model (note that we fix sampling 10,000 tasks from the meta-test set during final evaluation), default to200
-
lr
: Learning rate for the model, default to0.001
with pre-trained weights -
lr_mul
: This is specially designed for set-to-set functions like FEAT. The learning rate for the top layer will be multiplied by this value (usually with faster learning rate). Default to10
-
lr_scheduler
: The scheduler to set the learning rate (step
,multistep
, orcosine
), default tostep
-
step_size
: The step scheduler to decrease the learning rate. Set it to a single value if choose thestep
scheduler and provide multiple values when choosing themultistep
scheduler. Default to20
-
gamma
: Learning rate ratio forstep
ormultistep
scheduler, default to0.1
-
fix_BN
: Set the encoder to the evaluation mode during the meta-training. This parameter is useful when meta-learning with the WRN. Default toFalse
-
mom
: The momentum value for the SGD optimizer, default to0.9
-
weight_decay
: The weight_decay value for SGD optimizer, default to0.0005
Model Related Arguments
-
model_class
: The model to use during meta-learning. We provide implementations forMAML
and ourMAMLUnicorn
. Default toMAML
-
backbone_class
: Types of the encoder, i.e., ResNet-12 (Res12
), default toConvNet
-
temperature
: Temperature over the logits, we #divide# logits with this value. It is useful when meta-learning with pre-trained weights. Default to0.5
Other Arguments
-
gpu
: The index of GPU to use. Please provide multiple indexes if choosemulti_gpu
. Default to0
-
log_interval
: How often to log the meta-training information, default to every50
tasks -
eval_interval
: How often to validate the model over the meta-val set, default to every1
epoch -
save_dir
: The path to save the learned models, default to./checkpoints
Running the command without arguments will train the models with the default hyper-parameter values. Loss changes will be recorded as a tensorboard file.
For example, to train the 1-shot/5-shot 5-way MAML/UNICORN-MAML model with ResNet-12 backbone on MiniImageNet:
$ python train_fsl.py --max_epoch 100 --way 5 --eval_way 5 --lr_scheduler step --model_class MAML --lr_mul 10 --backbone_class Res12 --dataset MiniImageNet --gpu 0 --query 15 --step_size 20 --gamma 0.1 --para_init './saves/initialization/miniimagenet/Res12-pre.pth' --lr 0.001 --shot 1 --eval_shot 1 --temperature 0.5 --gd_lr 0.05 --inner_iters 15
$ python train_fsl.py --max_epoch 100 --way 5 --eval_way 5 --lr_scheduler step --model_class MAML --lr_mul 10 --backbone_class Res12 --dataset MiniImageNet --gpu 0 --query 15 --step_size 20 --gamma 0.1 --para_init './saves/initialization/miniimagenet/Res12-pre.pth' --lr 0.001 --shot 5 --eval_shot 5 --temperature 0.5 --gd_lr 0.1 --inner_iters 20
$ python train_fsl.py --max_epoch 100 --way 5 --eval_way 5 --lr_scheduler step --model_class MAMLUnicorn --lr_mul 10 --backbone_class Res12 --dataset MiniImageNet --gpu 0 --query 15 --step_size 20 --gamma 0.1 --para_init './saves/initialization/miniimagenet/Res12-pre.pth' --lr 0.001 --shot 1 --eval_shot 1 --temperature 0.5 --gd_lr 0.1 --inner_iters 5
$ python train_fsl.py --max_epoch 100 --way 5 --eval_way 5 --lr_scheduler step --model_class MAMLUnicorn --lr_mul 10 --backbone_class Res12 --dataset MiniImageNet --gpu 0 --query 15 --step_size 20 --gamma 0.1 --para_init './saves/initialization/miniimagenet/Res12-pre.pth' --lr 0.001 --shot 5 --eval_shot 5 --temperature 0.5 --gd_lr 0.1 --inner_iters 20
to train the 1-shot/5-shot 5-way MAML/UNICORN-MAML model with ResNet-12 backbone on TieredImageNet:
$ python train_fsl.py --max_epoch 100 --way 5 --eval_way 5 --lr_scheduler step --model_class MAML --lr_mul 10 --backbone_class Res12 --dataset TieredImageNet --gpu 0 --query 15 --step_size 20 --gamma 0.1 --para_init './saves/initialization/tieredimagenet/Res12-pre.pth' --lr 0.001 --shot 1 --eval_shot 1 --temperature 0.5 --gd_lr 0.01 --inner_iters 20
$ python train_fsl.py --max_epoch 100 --way 5 --eval_way 5 --lr_scheduler step --model_class MAML --lr_mul 10 --backbone_class Res12 --dataset TieredImageNet --gpu 0 --query 15 --step_size 20 --gamma 0.1 --para_init './saves/initialization/tieredimagenet/Res12-pre.pth' --lr 0.001 --shot 1 --eval_shot 5 --temperature 0.5 --gd_lr 0.05 --inner_iters 15
$ python train_fsl.py --max_epoch 100 --way 5 --eval_way 5 --lr_scheduler step --model_class MAMLUnicorn --lr_mul 10 --backbone_class Res12 --dataset TieredImageNet --gpu 0 --query 15 --step_size 20 --gamma 0.1 --para_init './saves/initialization/tieredimagenet/Res12-pre.pth' --lr 0.001 --shot 5 --eval_shot 1 --temperature 0.5 --gd_lr 0.02 --inner_iters 10
$ python train_fsl.py --max_epoch 100 --way 5 --eval_way 5 --lr_scheduler step --model_class MAMLUnicorn --lr_mul 10 --backbone_class Res12 --dataset TieredImageNet --gpu 0 --query 15 --step_size 20 --gamma 0.1 --para_init './saves/initialization/tieredimagenet/Res12-pre.pth' --lr 0.001 --shot 1 --eval_shot 5 --temperature 0.5 --gd_lr 0.05 --inner_iters 20
We can evaluate a learned MAML model and check whether the permutation will introduce large variance. For example, 1-shot/5-shot 5-way model with ResNet-12 backbone on MiniImageNet:
$ python eval_maml_permutation.py --shot_list 1 --model_path './MAML-1-shot.pth' --gpu 0 --gd_lr 0.05 --inner_iters 15 --model_class MAML --dataset MiniImageNet
$ python eval_maml_permutation.py --shot_list 5 --model_path './MAML-5-shot.pth' --gpu 0 --gd_lr 0.1 --inner_iters 20 --model_class MAML --dataset MiniImageNet
We thank the following repos providing helpful components/functions in our work.