Dirichlet-based Uncertainty Calibration for Active Domain Adaptation
Authors: Mixue Xie, Shuang Li, Rui Zhang, Chi Harold Liu
ICLR 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments on cross-domain image classification and semantic segmentation validate the superiority of DUC. |
| Researcher Affiliation | Academia | Mixue Xie, Shuang Li , Rui Zhang, Chi Harold Liu Beijing Institute of Technology, China |
| Pseudocode | Yes | B. Algorithm of DUC Algorithm 1 Pseudo code of the proposed DUC |
| Open Source Code | Yes | Code is available at https://github.com/BIT-DA/DUC. |
| Open Datasets | Yes | We evaluate DUC on three cross-domain image classification datasets: mini Domain Net (Zhou et al., 2021), Office-Home (Venkateswara et al., 2017), Vis DA-2017 (Peng et al., 2017), and two adaptive semantic segmentation tasks: GTAV (Richter et al., 2016) Cityscapes (Cordts et al., 2016), SYNTHIA (Ros et al., 2016) Cityscapes. |
| Dataset Splits | Yes | Cityscapes (Cordts et al., 2016) gathers 5,000 images of urban street scenes from real world... These images are divided into training, validation and test splits. Similar to (Ning et al., 2021), we use the training split with 2,975 images as target training data, where labels are not used, and the model is evaluated on the validation split with 500 images by reporting the m Io U of the common categories. |
| Hardware Specification | Yes | We run each task on a single NVIDIA Ge Force RTX 2080 Ti GPU." and "For each semantic segmentation task, we run the experiment on a single NVIDIA Ge Force RTX 3090 GPU. |
| Software Dependencies | No | All experiments are implemented via Py Torch (Paszke et al., 2019). While PyTorch is mentioned, specific version numbers for PyTorch or other software dependencies like Python or CUDA are not provided. |
| Experiment Setup | Yes | We adopt the mini-batch SGD optimizer with batch size 32, momentum 0.9 to optimize the model. As for hyper-parameters, we select them by the grid search with deep embedded validation (DEV) (You et al., 2019) and use β = 1.0, λ = 0.05, κ = 10 for image classification. ... Similarly, the mini-batch SGD optimizer is adopted, where batch size is 2. And we set β = 1.0, λ = 0.01, κ = 10 for semantic segmentation. For all tasks, we report the mean std of 3 random trials... |