Discriminative Feature Generation for Classification of Imbalanced Data in PyTorch

A Pytorch implementation of Discriminative feature generation for classification of imbalanced data described in the paper:

• Sungho Suh, Paul Lukowicz, and Yong Oh Lee, "Discriminative feature generation for classification of imbalanced data", Pattern Recognition, 2022. [Pattern Recognition] [arXiv]

Abstract

The data imbalance problem is frequently bottleneck of the neural network performance in classification. In this paper, we propose a novel supervised discriminative feature generation method (DFG) for minority class dataset. DFG is based on the modified structure of Generative Adversarial Network consisting of four independent networks: generator, discriminator, feature extractor, and classifier. To augment the selected discriminative features of minority class data by adopting attention mechanism, the generator for class-imbalanced target task is trained while feature extractor and classifier are regularized with the pre-trained ones from large source data. The experimental results show that the generator of DFG enhances the augmentation of label-preserved and diverse features, and classification results are significantly improved on the target task.

Models

The performance of each model

		LeNet-5 (EMNIST)		VGG-16 (CIFAR10)		ResNet-50 (ImageNet)
Data Set		SVHN	F-MNIST	STL-10	CINIC-10	CALTECH-256	FOOD-101
Imbalance ratio (IR)		10:1	40:1	1:1	10:1	1:1	5:1
ORIGINAL		76.57 ± 1.65	77.13 ± 3.24	66.73 ± 0.67	58.14 ± 3.42	43.30 ± 0.34	30.17 ± 1.72
FINE-TUNING		76.66 ± 0.65	75.39 ± 3.54	72.89 ± 0.38	64.42 ± 0.81	81.99 ± 0.12	68.99 ± 0.42
DELTA		78.47 ± 0.57	78.91 ± 2.27	79.41 ± 0.27	68.89 ± 0.18	85.33 ± 0.15	72.03 ± 0.35
DIFA + CMWGAN		76.64 ± 1.32	78.27 ± 2.32	80.08 ± 0.19	71.82 ± 0.35	82.23 ± 0.32	71.71 ± 0.08
OURS(DFG)		80.81 ± 0.25	82.65 ± 0.28	81.09 ± 0.17	72.09 ± 0.20	86.29 ± 0.19	76.00 ± 0.36

Prerequisites

• Linux (Ubuntu)
• Python >= 3.6
• NVIDIA GPU + CUDA CuDNN

Installation

• Clone this repo:

git clone https://github.com/opensuh/DFG/

• Install PyTorch
  • For pip users, please type the command pip install -r requirements.txt.
  • For Conda users, you can create a new Conda environment using conda env create -f environment.yml.

DFG train/eval

• Prepare datasets

  • SVHN
  • F-MNIST
  • STL-10
  • CINIC-10
  • CALTECH-256
  • FOOD-101

• Train a model:

./scripts/lenet_svhn_train.sh
./scripts/lenet_fmnist_train.sh
./scripts/vgg_stl_train.sh
./scripts/vgg_cinic_train.sh
./scripts/resnet_caltech_train.sh
./scripts/resnet_food_train.sh

• Evaluate the model (our pre-trained models are in ./pretrained_model)
• We plan to upload the pre-trained models on our Github page.

./scripts/lenet_svhn_eval.sh
./scripts/lenet_fmnist_eval.sh
./scripts/vgg_stl_eval.sh
./scripts/vgg_cinic_eval.sh
./scripts/resnet_caltech_eval.sh
./scripts/resnet_food_eval.sh