Deep Asymmetric Transfer Network for Unbalanced Domain Adaptation

Authors: Daixin Wang, Peng Cui, Wenwu Zhu

AAAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The experiments on two real-world datasets demonstrate that DATN attains a substantial gain over state-of-the-art methods.
Researcher Affiliation Academia Daixin Wang, Peng Cui, Wenwu Zhu Tsinghua National Laboratory for Information Science and Technology, Department of Computer Science and Technology, Tsinghua University, Beijing, China dxwang0826@gmail.com, cuip@mail.tsinghua.edu.cn, wwzhu@tsinghua.edu.cn
Pseudocode Yes The full algorithm is shown in Alg. 1.
Open Source Code No The paper does not provide any statement or link regarding the release of open-source code for the described methodology.
Open Datasets Yes In our experiments, we use two real-world datasets, i.e. NUS-WIDE and AMAZON REVIEWS. NUS-WIDE (Chua et al. 2009) is a public web image dataset... AMAZON REVIEWS (Prettenhofer and Stein 2010) is a cross-language dataset...
Dataset Splits Yes The final values of all the parameters are determined by using 5-fold cross-validation on the training set.
Hardware Specification No The paper states 'Our approach is implemented in Tensorflow' but does not provide any specific details about the hardware (e.g., GPU, CPU models) used for running the experiments.
Software Dependencies No The paper mentions 'Our approach is implemented in Tensorflow' but does not specify a version number for TensorFlow or any other software dependencies.
Experiment Setup Yes The values of α and β are selected from {0, 0.5, 1, 2, 5, 10}. The regularization parameters of λ and λ are set as 0.1 and 0.0001. Throughout the experiments, the learning rate is set as 0.0001, the decay is set as 0.8 and the momentum is set as 0.8. Table 3: Number of neurons of each layer of DATN.