Rethinking Task-Incremental Learning Baselines

Official imlpementation of the paper [ICPR 2022]

Link to the Paper - arXiv

Prerequisites:

• Linux-64
• Python 3.9
• PyTorch 1.10.1
• CPU or NVIDIA GPU + CUDA10.2 CuDNN7.5

Installation

• Create a conda environment and install required packages:

conda create -n <env> python=3.9
conda activate <env>
pip install -r requirements.txt

Datasets

Download the Mini-imagenet and notMNIST datasets from Google Drive. Other datasets will be automatically downloaded.

Training

For training, run the following command.
python run.py -exp <experiment_id>

To manually input number of runs, epochs and learning rate, run the following command:
python run.py -exp <experiment_id> -r <n_runs> -e <n_epochs> -lr <learning_rate>

Test

For test, run the following command.
python test.py -exp <experiment_id>

Citation

If you wish to cite this project, feel free to use this BibTeX reference:

@misc{https://doi.org/10.48550/arxiv.2205.11367,
  doi = {10.48550/ARXIV.2205.11367},
  url = {https://arxiv.org/abs/2205.11367},
  author = {Hossain, Md Sazzad and Saha, Pritom and Chowdhury, Townim Faisal and Rahman, Shafin and Rahman, Fuad and Mohammed, Nabeel},
  keywords = {Artificial Intelligence (cs.AI), FOS: Computer and information sciences, FOS: Computer and information sciences},
  title = {Rethinking Task-Incremental Learning Baselines},
  publisher = {arXiv},
  year = {2022},
  copyright = {arXiv.org perpetual, non-exclusive license}
}