Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Learning Multi-Source and Robust Representations for Continual Learning
Authors: Fei Ye, Yongcheng Zhong, Qihe Liu, Adrian G. Bors, Jingling sun, Rongyao Hu, shijie zhou
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Our extensive experimental evaluations reveal that the proposed framework attains state-of-the-art performance. The source code of our algorithm is available at https://github.com/CL-Coder236/LMSRR. [...] 4 Experiment [...] 4.2 Results on Standard Datasets [...] 4.3 Results on Complex Datasets [...] 4.4 Ablation Study |
| Researcher Affiliation | Academia | 1School of Information and Software Engineering, University of Electronic Science and Technology of China 2Department of Computer Science, University of York {EMAIL, EMAIL, EMAIL, EMAIL, EMAIL, EMAIL, EMAIL} |
| Pseudocode | Yes | We provide the detailed learning process of the proposed framework in Fig. 1 while the detailed pseudocode is provided in Algorithm 1 which consists of three steps |
| Open Source Code | Yes | The source code of our algorithm is available at https://github.com/CL-Coder236/LMSRR. |
| Open Datasets | Yes | we conducted extensive experiments on seven different datasets, including CIFAR10 [33], Tiny Image Net [35], MNIST [36], CIFAR-100 [34], CUB-200 [55], Image Net-R [22], and Cars196 [32]. |
| Dataset Splits | No | In continual learning (CL), models face the limitation of being unable to access the entire training dataset. The training for each task is restricted to data samples pertinent to the current task, and data from previous tasks is inaccessible. A prominent scenario in this domain is Task-Incremental Learning (TIL), where the training dataset Ds = {(xj, yj) | j = 1, , N s} is divided into multiple taskspecific subsets {Ds 1, , Ds C }, each corresponding to an individual task Tj. [...] We provide the detailed experiment setting in Appendix A from Supplementary Material (SM). |
| Hardware Specification | No | We provide the detailed experiment setting in Appendix A from Supplementary Material (SM). |
| Software Dependencies | No | The main paper text does not explicitly mention any specific software dependencies with version numbers. It states, "We provide the algorithm implementation in Section 3.6, along with the source code. The detailed experimental setup is documented in Appendix-A.", implying these details are in supplementary materials, not the main body. |
| Experiment Setup | No | We provide the detailed experiment setting in Appendix A from Supplementary Material (SM). |