Few-Shot Learning via Embedding Adaptation with Set-to-Set Functions

The code repository for "Few-Shot Learning via Embedding Adaptation with Set-to-Set Functions" [paper] [ArXiv] [slides] [poster] (Accepted by CVPR 2020) in PyTorch. If you use any content of this repo for your work, please cite the following bib entry:

@inproceedings{ye2020fewshot,
  author    = {Han-Jia Ye and
               Hexiang Hu and
               De-Chuan Zhan and
               Fei Sha},
  title     = {Few-Shot Learning via Embedding Adaptation with Set-to-Set Functions},
  booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  pages     = {8808--8817},
  year      = {2020}
}

Embedding Adaptation with Set-to-Set Functions

We propose a novel model-based approach to adapt the instance embeddings to the target classification task with a #set-to-set# function, yielding embeddings that are task-specific and are discriminative. We empirically investigated various instantiations of such set-to-set functions and observed the Transformer is most effective --- as it naturally satisfies key properties of our desired model. We denote our method as Few-shot Embedding Adaptation with Transformer (FEAT).

Standard Few-shot Learning Results

Experimental results on few-shot learning datasets with ResNet-12 backbone (Same as this repo). 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	Link to Weights
ProtoNet	62.39	80.53	1-Shot, 5-Shot
BILSTM	63.90	80.63	1-Shot, 5-Shot
DEEPSETS	64.14	80.93	1-Shot, 5-Shot
GCN	64.50	81.65	1-Shot, 5-Shot
FEAT	66.78	82.05	1-Shot, 5-Shot

TieredImageNet Dataset

Setups	1-Shot 5-Way	5-Shot 5-Way	Link to Weights
ProtoNet	68.23	84.03	1-Shot, 5-Shot
BILSTM	68.14	84.23	1-Shot, 5-Shot
DEEPSETS	68.59	84.36	1-Shot, 5-Shot
GCN	68.20	84.64	1-Shot, 5-Shot
FEAT	70.80	84.79	1-Shot, 5-Shot

Prerequisites

The following packages are required to run the scripts:

PyTorch-1.4 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: please download the pre-trained weights of the encoder if needed. The pre-trained weights can be downloaded in a zip file.

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 SEEN categories, 16 and 20 as two sets of UNSEEN categories for model validation and evaluation, respectively.

CUB Dataset

Caltech-UCSD Birds (CUB) 200-2011 dataset is initially designed for fine-grained classification. It contains in total 11,788 images of birds over 200 species. On CUB, we randomly sampled 100 species as SEEN classes, and another two 50 species are used as two UNSEEN sets. We crop all images with given bounding boxes before training. We only test CUB with the ConvNet backbone in our work.

TieredImageNet Dataset

TieredImageNet is a large-scale dataset with more categories, which contains 351, 97, and 160 categoriesfor model training, validation, and evaluation, respectively. The dataset can also be download from here. We only test TieredImageNet with ResNet backbone in our work.

Check this for details of data downloading and preprocessing.

Code Structures

To reproduce our experiments with FEAT, 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. FEAT meta-learns the embedding adaptation process such that all the training instance embeddings in a task is adapted, based on their contextual task information, using Transformer. 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 0.0001 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 0.2
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
augment: Whether to do data augmentation or not during meta-training, 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 baselines (MatchNet and ProtoNet), set-to-set functions (BILSTM, DeepSet, GCN, and our FEAT variants). We also include an instance-specific embedding adaptation approach FEAT, which is discussed in the old version of the paper. SemiFEAT is the one which combines the unlabeled query set instances into the feature adaptation in a transductive manner, while SemiProtoFEAT applies Semi-ProtoNet over the transductively transformed embeddings of SemiFEAT. Default to FEAT
use_euclidean: Use the euclidean distance or the cosine similarity to compute pairwise distances. We use the euclidean distance in the paper. Default to False
backbone_class: Types of the encoder, i.e., the convolution network (ConvNet), ResNet-12 (Res12), or Wide ResNet (WRN), default to ConvNet
balance: This is the balance weight for the contrastive regularizer. Default to 0
temperature: Temperature over the logits, we #divide# logits with this value. It is useful when meta-learning with pre-trained weights. Default to 1
temperature2: Temperature over the logits in the regularizer, we divide logits with this value. This is specially designed for the contrastive regularizer. Default to 1

Other Arguments

orig_imsize: Whether to resize the images before loading the data into the memory. -1 means we do not resize the images and do not read all images into the memory. Default to -1
multi_gpu: Whether to use multiple gpus during meta-training, default to False
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 FEAT

For example, to train the 1-shot/5-shot 5-way FEAT model with ConvNet backbone on MiniImageNet:

$ python train_fsl.py  --max_epoch 200 --model_class FEAT --use_euclidean --backbone_class ConvNet --dataset MiniImageNet --way 5 --eval_way 5 --shot 1 --eval_shot 1 --query 15 --eval_query 15 --balance 1 --temperature 64 --temperature2 16 --lr 0.0001 --lr_mul 10 --lr_scheduler step --step_size 20 --gamma 0.5 --gpu 8 --init_weights ./saves/initialization/miniimagenet/con-pre.pth --eval_interval 1
$ python train_fsl.py  --max_epoch 200 --model_class FEAT --use_euclidean --backbone_class ConvNet --dataset MiniImageNet --way 5 --eval_way 5 --shot 5 --eval_shot 5 --query 15 --eval_query 15 --balance 0.1 --temperature 32 --temperature2 64 --lr 0.0001 --lr_mul 10 --lr_scheduler step --step_size 20 --gamma 0.5 --gpu 14 --init_weights ./saves/initialization/miniimagenet/con-pre.pth --eval_interval 1

to train the 1-shot/5-shot 5-way FEAT model with ResNet-12 backbone on MiniImageNet:

$ python train_fsl.py  --max_epoch 200 --model_class FEAT  --backbone_class Res12 --dataset MiniImageNet --way 5 --eval_way 5 --shot 1 --eval_shot 1 --query 15 --eval_query 15 --balance 0.01 --temperature 64 --temperature2 64 --lr 0.0002 --lr_mul 10 --lr_scheduler step --step_size 40 --gamma 0.5 --gpu 1 --init_weights ./saves/initialization/miniimagenet/Res12-pre.pth --eval_interval 1 --use_euclidean
$ python train_fsl.py  --max_epoch 200 --model_class FEAT  --backbone_class Res12 --dataset MiniImageNet --way 5 --eval_way 5 --shot 5 --eval_shot 5 --query 15 --eval_query 15 --balance 0.1 --temperature 64 --temperature2 32 --lr 0.0002 --lr_mul 10 --lr_scheduler step --step_size 40 --gamma 0.5 --gpu 0 --init_weights ./saves/initialization/miniimagenet/Res12-pre.pth --eval_interval 1 --use_euclidean

to train the 1-shot/5-shot 5-way FEAT model with ResNet-12 backbone on TieredImageNet:

$ python train_fsl.py  --max_epoch 200 --model_class FEAT  --backbone_class Res12 --dataset TieredImageNet --way 5 --eval_way 5 --shot 1 --eval_shot 1 --query 15 --eval_query 15 --balance 0.1 --temperature 64 --temperature2 64 --lr 0.0002 --lr_mul 10 --lr_scheduler step --step_size 20 --gamma 0.5 --gpu 0 --init_weights ./saves/initialization/tieredimagenet/Res12-pre.pth --eval_interval 1  --use_euclidean
$ python train_fsl.py  --max_epoch 200 --model_class FEAT  --backbone_class Res12 --dataset TieredImageNet --way 5 --eval_way 5 --shot 5 --eval_shot 5 --query 15 --eval_query 15 --balance 0.1 --temperature 32 --temperature2 64 --lr 0.0002 --lr_mul 10 --lr_scheduler step --step_size 40 --gamma 0.5 --gpu 0 --init_weights ./saves/initialization/tieredimagenet/Res12-pre.pth --eval_interval 1  --use_euclidean

Acknowledgment

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

junayed/FEAT