Domain Adaptive Classification on Heterogeneous Information Networks

Authors: Shuwen Yang, Guojie Song, Yilun Jin, Lun Du

IJCAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments on pairwise datasets endorse not only our model s performance on domain adaptive classification on HINs and contributions by individual components. We evaluate Mu SDAC quantitatively on three pairs of networks where Mu SDAC outperforms various baselines on transferable classification. We also carry out model analysis and visualization to verify contributions of individual model components.
Researcher Affiliation Collaboration Shuwen Yang1 , Guojie Song1 , Yilun Jin2 and Lun Du3 1Key Laboratory of Machine Perception (Ministry of Education), Peking University, China 2The Hong Kong University of Science and Technology, Hong Kong SAR, China 3Microsoft Research, China {swyang, gjsong}@pku.edu.cn, yilun.jin@connect.ust.hk, lun.du@microsoft.com
Pseudocode Yes Algorithm 1 Heuristic Combination Sampling Algorithm
Open Source Code Yes Code available on https://github.com/PKUterran/Mu SDAC
Open Datasets Yes Datasets We sample pairs of structurally different graphs respectively from ACM [Kong et al., 2012], AMiner and DBLP [Wang et al., 2019]. The explicit description of datasets is on https://github.com/PKUterran/Mu SDAC/blob/master/data/DATA.md.
Dataset Splits No The paper mentions using 'ACM', 'AMiner', and 'DBLP' datasets but does not provide specific details about train/validation/test splits (e.g., percentages, sample counts, or a standard split reference within the paper's main text).
Hardware Specification No The paper does not provide any specific hardware details such as CPU/GPU models, memory, or other compute infrastructure used for the experiments.
Software Dependencies No The paper does not list any specific software dependencies with version numbers (e.g., Python 3.x, PyTorch 1.x, CUDA x.x).
Experiment Setup Yes In Mu SDAC and its variants, the dimensionality of the first and second hidden layers of the multichannel GCN is 64 and 32 respectively, before aggregated to 16 in the aggregated channel. The number of sampled combinations |Z| = M = 2N 1. In DAC, we use 5 Gaussian kernels for MMD and γ = 10. In weighted voting, we take η = 25 and α = 0.95.