Keypoint-Guided Optimal Transport with Applications in Heterogeneous Domain Adaptation

Authors: Xiang Gu, Yucheng Yang, Wei Zeng, Jian Sun, Zongben Xu

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments verified the effectiveness of the KPG-RL model. Code is available at https://github.com/XJTU-XGU/KPG-RL.
Researcher Affiliation Academia Xiang Gu1, Yucheng Yang1, Wei Zeng1, Jian Sun ( )123, Zongben Xu123 1 School of Mathematics and Statistics, Xi an Jiaotong University, Xi an, China 2 Pazhou Laboratory (Huangpu), Guangzhou, China 3 Peng Cheng Laboratory, Shenzhen, China {xianggu,ycyang}@stu.xjtu.edu.cn {wz,jiansun,zbxu}@xjtu.edu.cn
Pseudocode No The paper describes iterative formulas and steps for its model but does not include a formally labeled 'Algorithm' or 'Pseudocode' block.
Open Source Code Yes Code is available at https://github.com/XJTU-XGU/KPG-RL.
Open Datasets Yes We conduct experiments on Office-31 [53] dataset.
Dataset Splits No The paper specifies training data (transported source, labeled target, unlabeled target) and test data, but does not explicitly describe a separate validation split or how it was used.
Hardware Specification No The experiments are conducted on CPU. No specific CPU model, GPU, or other hardware details are provided.
Software Dependencies No The paper does not mention specific software dependencies with version numbers (e.g., Python, PyTorch versions, or specific solvers).
Experiment Setup Yes ϵ is set to 0.005. ρ is a tunable parameter and set to 0.1 in our experiments, since 0.1 is a commonly used temperature in the softmax function [47, 48]. α is simply set to 0.5 in our experiments. We train the classification model (taken as a kernel SVM).