A Model or 603 Exemplars: Towards Memory-Efficient Class-Incremental Learning

Authors: Da-Wei Zhou, Qi-Wei Wang, Han-Jia Ye, De-Chuan Zhan

ICLR 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments on benchmark datasets validate MEMO s competitive performance.
Researcher Affiliation Academia State Key Laboratory for Novel Software Technology, Nanjing University {zhoudw, wangqiwei, yehj, zhandc}@lamda.nju.edu.cn
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks.
Open Source Code Yes Code is available at: https://github.com/wangkiw/ICLR23-MEMO
Open Datasets Yes we evaluate the performance on CIFAR100 (Krizhevsky et al., 2009), and Image Net100/1000 (Deng et al., 2009).
Dataset Splits Yes CIFAR100 contains 50,000 training and 10,000 testing images, with a total of 100 classes. Image Net is a large-scale dataset with 1,000 classes, with about 1.28 million images for training and 50,000 for validation. The class order of training classes is shuffled with random seed 1993.
Hardware Specification Yes All models are deployed with Py Torch (Paszke et al., 2019) and Py CIL (Zhou et al., 2021a) on NVIDIA 3090.
Software Dependencies No The paper mentions PyTorch and PyCIL with citations but does not provide specific version numbers for these software dependencies.
Experiment Setup Yes The model is trained with a batch size of 128 for 170 epochs, and we use SGD with momentum for optimization. The learning rate starts from 0.1 and decays by 0.1 at 80 and 150 epochs.