Progressive Mix-Up for Few-Shot Supervised Multi-Source Domain Transfer
Authors: Ronghang Zhu, Ronghang Zhu, Xiang Yu, Sheng Li
ICLR 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments on two benchmarks show that our P-Mixup significantly outperforms the state-of-the-art methods, i.e., 6.0% and 8.6% improvements on Office-Home and Domain Net. |
| Researcher Affiliation | Collaboration | Ronghang Zhu1, Xiang Yu2, Sheng Li3,1 1University of Georgia, 2Amazon, 3University of Virginia |
| Pseudocode | No | The paper does not include pseudocode or clearly labeled algorithm blocks. It describes the architecture and learning objectives using mathematical formulas and diagrams. |
| Open Source Code | Yes | Source code is available at https://github.com/ronghangzhu/P-Mixup |
| Open Datasets | Yes | We adopt two standard domain adaptation and generalization benchmarks: (1) Office Home (Venkateswara et al., 2017) which consists of four domains (Art, Clipart, Produce, and Real world) with 65 classes. (2) Domain Net (Peng et al., 2019) contains 345 classes. |
| Dataset Splits | Yes | To exactly constrain the target distribution to be few-shot, for Office-Home, we randomly select 10 out of 65 classes each with one sample as the target. Similarly, we randomly select 15 out of 345 classes for Domain Net. The remaining samples in these selected classes are used as the test data. Such random sampling is conducted for 5 times and the averaged result is reported. ... Following MAML, we conduct a meta-optimization to pseudo-update the model parameters for the first time by minimizing P Ti p(T ) LTi ce: ... For each 3 iterations, we randomly select 2 source domains as the meta-train domain, and the rest source, target, and mixup domains are meta-test domain. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., CPU, GPU models, or memory) used for running its experiments, only mentioning the use of a pre-trained ResNet-18 backbone. |
| Software Dependencies | No | Our implementation is based on Pytorch (Paszke et al., 2019). We use Res Net-18 (He et al., 2016) pretrained on Image Net (Deng et al., 2009) as the backbone network. |
| Experiment Setup | Yes | Our implementation is based on Pytorch (Paszke et al., 2019). We use Res Net-18 (He et al., 2016) pretrained on Image Net (Deng et al., 2009) as the backbone network. We optimize the model using SGD with momentum of 0.9 and weight decay of 5 10 4. The batch size is set to 50. The initial learning rate is set to 0.001. ... We train the model for 10,000 iterations on Office-Home, and 30,000 iterations on Domain Net. We update the mix-up ratio λn every 100 iterations. ... The learning rates α and β are set to 0.01. ... σ is a local perturbation range, i.e., we empirically set it as 0.2. |