Online Boundary-Free Continual Learning by Scheduled Data Prior

Authors: Hyunseo Koh, Minhyuk Seo, Jihwan Bang, Hwanjun Song, Deokki Hong, Seulki Park, Jung-Woo Ha, Jonghyun Choi

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

Reproducibility Variable Result LLM Response
Research Type Experimental We empirically evaluate our method on a Gaussian data stream and its periodic extension, which is frequently observed in real-life data, as well as the conventional disjoint task-split. Our method outperforms prior arts by large margins in various setups, using four benchmark datasets in continual learning literature CIFAR-10, CIFAR-100, Tiny Image Net and Image Net.
Researcher Affiliation Collaboration 1Yonsei University 2GIST 3NAVER AI Lab 4NAVER Cloud 5Seoul National University
Pseudocode Yes Algorithm 1 provides detailed pseudocode for SDP.
Open Source Code Yes Code is available at https://github.com/yonseivnl/sdp.
Open Datasets Yes For empirical validations, following (Koh et al., 2021; Guo et al., 2022), we use four benchmark datasets; CIFAR-10, CIFAR-100, Tiny Image Net and Image Net.
Dataset Splits No The paper mentions 'confidence on currently learned model at the current time step before using them for training, as a proxy of validation accuracy' in Section 5.2, but it does not provide specific details on a separate validation dataset split (e.g., percentages, sample counts, or predefined splits) for reproducibility.
Hardware Specification Yes We compare the computational complexity by reporting the training time on a single NVIDIA Ge Force RTX 2080Ti.
Software Dependencies No The paper mentions using 'Res Net-18' and 'Adam optimizer' but does not specify version numbers for any software dependencies (e.g., Python, PyTorch, TensorFlow, CUDA versions).
Experiment Setup Yes For CIFAR-10, CIFAR-100, Tiny Image Net and Image Net, we use batch size of 16, 16, 32, 256, number of updates per sample of 1, 3, 3, 0.25, memory size of 500, 2000, 4000, 20000, respectively. We use Adam optimizer with LR of 0.0003 for all datasets and setup.