Task-Free Continual Learning via Online Discrepancy Distance Learning

Authors: Fei Ye, Adrian G. Bors

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We perform several TFCL experiments with the proposed methodology, which demonstrate that the proposed approach achieves the state of the art performance. 5 Experiments We perform the experiments to address the following research questions: 1) What factors would cause the model s forgetting, and how to explain such behaviour? 2) How efficient is the proposed ODDL-S under TFCL benchmarks? 3) How important is each module in OODL-S?
Researcher Affiliation Academia Fei Ye and Adrian G. Bors Department of Computer Science University of York York, YO10 5GH, UK {fy689,adrian.bors}@york.ac.uk
Pseudocode Yes The algorithm has three main stages (pseudocode is provided in Algorithm 1, Appendix-E from SM)
Open Source Code Yes The code is available at https://dtuzi123.github.io/ODDL/.
Open Datasets Yes We adapt the TFCL setting from [8] which employs several datasets including Split MNIST [22, 53], Split CIFAR10 [18, 53] and Split CIFAR100 [18].
Dataset Splits No The detailed information for datasets, hyperparameters and network architectures is provided in Appendix-F from the supplementary material. Specific training, validation, and test split percentages or counts are not explicitly stated in the main paper.
Hardware Specification No The paper does not explicitly describe the specific hardware used (e.g., GPU models, CPU types, or cloud instance specifications) for running experiments.
Software Dependencies No The paper does not provide specific software dependencies with version numbers (e.g., library names with explicit version details).
Experiment Setup Yes The hyperparameter λ used in Eq. (15) for expanding ODDL when learning Split MINI-Image Net and Permuted MNIST (where pixels are randomly premuted in images) is equal to 1.2 and 1.5, respectively. The detailed information for datasets, hyperparameters and network architectures is provided in Appendix-F from the supplementary material.