How to Train Your MAML to Excel in Few-Shot Classification

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}
}

Main idea of UNICORN-MAML

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.

Standard Few-shot Learning Results

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.

TieredImageNet Dataset

Prerequisites

The following packages are required to run the scripts:

• PyTorch-1.6 and torchvision

• 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.

Dataset

MiniImageNet Dataset

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 Dataset

TieredImageNet is a large-scale dataset with more categories, which contains 351, 97, and 160 categoriesfor model training, validation, and evaluation, respectively.

Code Structures

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.

Model Training and Evaluation

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.

Arguments

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, or CUB), default to MiniImageNet

• way: The number of classes in a few-shot task during meta-training, default to 5

• eval_way: The number of classes in a few-shot task during meta-test, default to 5

• shot: Number of instances in each class in a few-shot task during meta-training, default to 1

• eval_shot: Number of instances in each class in a few-shot task during meta-test, default to 1

• query: Number of instances in each class to evaluate the performance during meta-training, default to 15

• eval_query: Number of instances in each class to evaluate the performance during meta-test, default to 15

Optimization Related Arguments

• max_epoch: The maximum number of training epochs, default to 200

• episodes_per_epoch: The number of tasks sampled in each epoch, default to 100

• 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 to 200

• lr: Learning rate for the model, default to 0.001 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 to 10

• lr_scheduler: The scheduler to set the learning rate (step, multistep, or cosine), default to step

• step_size: The step scheduler to decrease the learning rate. Set it to a single value if choose the step scheduler and provide multiple values when choosing the multistep scheduler. Default to 20

• gamma: Learning rate ratio for step or multistep scheduler, default to 0.1

• fix_BN: Set the encoder to the evaluation mode during the meta-training. This parameter is useful when meta-learning with the WRN. Default to False

• mom: The momentum value for the SGD optimizer, default to 0.9

• weight_decay: The weight_decay value for SGD optimizer, default to 0.0005

Model Related Arguments

• model_class: The model to use during meta-learning. We provide implementations for MAML and ourMAMLUnicorn. Default to MAML

• backbone_class: Types of the encoder, i.e., ResNet-12 (Res12), default to ConvNet

• temperature: Temperature over the logits, we #divide# logits with this value. It is useful when meta-learning with pre-trained weights. Default to 0.5

Other Arguments

• gpu: The index of GPU to use. Please provide multiple indexes if choose multi_gpu. Default to 0

• log_interval: How often to log the meta-training information, default to every 50 tasks

• eval_interval: How often to validate the model over the meta-val set, default to every 1 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.

Training scripts for UNICORN-MAML

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 

Verifying the permutation variance of a learned MAML model

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

Acknowledgment

We thank the following repos providing helpful components/functions in our work.

• FEAT

• AVIATOR