Skip to content

Latest commit

 

History

History
54 lines (42 loc) · 2.08 KB

README.md

File metadata and controls

54 lines (42 loc) · 2.08 KB
Raw

DRT : Dynamic Replay Training for class-incremental learning

Introduction • Usage • Acknowledgments • Contact •

Created by Yan Yang from Nanjing University.
This project is built upon the PyCIL library developed by Da-Wei Zhou.
Link to the PyCIL library: PyCIL: A Python Toolbox for Class-Incremental Learning

Introduction

DRT aims to reduce catastrophic forgetting caused by class imbalance in class-incremental learning by dynamically replaying the memory data.
A more detailed abstract can be found in the paper. This code implements our method. We use Python and Pytorch for the experiments.

Usage

The code has been tested on Linux (Linux version 3.10.0-1160.25.1.el7.x86_64)

Dependencies

  1. torch 1.81
  2. torchvision 0.6.0
  3. tqdm
  4. numpy
  5. scipy
  6. quadprog

Run experiment

  1. Edit the exps/[MODEL NAME].json file for global settings.
  2. As for our method, you can run:
python main.py --config=./exps/drt.json

Datasets

The code has implemented the pre-processing of CIFAR100 and imagenet100. When training on CIFAR100, this framework will automatically download it. When training on imagenet100, you should specify the folder of your dataset in utils/data.py.

    def download_data(self):
        assert 0,"You should specify the folder of your dataset"
        train_dir = '[DATA-PATH]/train/'
        test_dir = '[DATA-PATH]/val/'

Acknowledgments

We appreciate the open source code framework provided by the PyCIL project.

Contact

If there are any questions, please feel free to propose new features by opening an issue or contact with the author: Yan Yang (yangyan@smail.nju.edu.cn). Enjoy the code.