CD^2: Constrained Dataset Distillation for Few-Shot Class-Incremental Learning
Authors: Kexin Bao, Daichi Zhang, Hansong Zhang, Yong Li, Yutao Yue, Shiming Ge
IJCAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments on three public datasets show the superiority of our method against other state-of-the-art competitors. |
| Researcher Affiliation | Academia | 1Institute of Information Engineering, Chinese Academy of Sciences 2School of Cyber Security, University of Chinese Academy of Sciences 3Hong Kong University of Science and Technology (Guangzhou) EMAIL, EMAIL, EMAIL |
| Pseudocode | No | The paper describes the methodology in prose and mathematical equations but does not include any explicitly labeled pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not contain any explicit statements or links indicating the public availability of source code for the described methodology. |
| Open Datasets | Yes | We evaluate our CD2 on three FSCIL benchmark datasets: CIFAR100 [Krizhevsky and Hinton, 2009], mini Image Net [Russakovsky et al., 2015], and CUB200 [Wah et al., 2011] following previous settings [Liu et al., 2022; Yang et al., 2023b]. |
| Dataset Splits | Yes | On CIFAR100 and mini-Image Net, the 100 classes are organized into 60 base classes and 40 incremental classes. 60 base classes contain 500 training images for the training model. And 40 incremental classes are further structured in 8 different sets with a 5-way 5-shot setting. And 200 classes of CUB200 are organized into 100 base classes and 100 incremental classes in a 10-way 5-shot FSCIL setting for 10 incremental sessions. |
| Hardware Specification | No | The paper mentions the use of ResNet backbones (ResNet12 and ResNet18) but does not provide any specific details about the hardware (GPU, CPU, memory, etc.) used for experiments. |
| Software Dependencies | No | The paper mentions optimizers (SGD) and data augmentation techniques (Mixup, CutMix) but does not specify programming languages, deep learning frameworks, or other software dependencies with version numbers. |
| Experiment Setup | Yes | Our model is optimized using SGD with momentum and adopts a cosine annealing strategy for the learning rate during training. In the base session, we train for 100 to 200 epochs while initializing a learning rate of 0.25 for CIFAR100 and mini-Image Net, and 0.01 for CUB200. In each incremental session, we train for 100 to 300 iterations initializing a learning rate of 0.25 for CIFAR100 and Mini Image Net, and 0.001 for CUB200. Augmentations include random resizing, random flipping, Mixup [Zhang et al., 2018], and Cut Mix [Yun et al., 2019]. And we set β = 0.1 for model training. When using DDM, we train 1000 iterations initializing a learning rate of 0.2. |