Belief Propagation Network for Hard Inductive Semi-Supervised Learning

Authors: Jaemin Yoo, Hyunsik Jeon, U Kang

IJCAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We conduct extensive experiments to demonstrate the superior performance of BPN, which shows the highest classification accuracy on four datasets compared with the state-of-the-art approaches for inductive learning.
Researcher Affiliation Academia Jaemin Yoo , Hyunsik Jeon and U Kang Seoul National University {jaeminyoo, jeon185, ukang}@snu.ac.kr
Pseudocode Yes Algorithm 1 Belief Propagation Network (BPN)
Open Source Code No The paper mentions obtaining 'public implementations of the baseline methods' and provides a GitHub link for Planetoid (a baseline method), but does not state that the source code for BPN is publicly available or provided.
Open Datasets Yes We use four datasets summarized in Table 2. The first three datasets [Sen et al., 2008] were used to evaluate the previous approaches [Velickovic et al., 2018]. ... We also use an Amazon dataset based on [Mc Auley et al., 2015; He and Mc Auley, 2016].
Dataset Splits Yes For each dataset, we use 20 nodes of each class for training, 1,000 nodes for testing, and 500 nodes for validation as done in [Kipf and Welling, 2017].
Hardware Specification Yes Our experiments are done in a workstation with Geforce GTX 1080 Ti.
Software Dependencies No The paper mentions using 'a recent deep learning framework' and 'Adam' as an optimizer but does not specify software names with version numbers (e.g., Python 3.x, TensorFlow 2.x, PyTorch 1.x).
Experiment Setup Yes We use a feedforward neural network with one hidden layer as a classifier f. ... The number of hidden units is set to 32, and we use dropout [Srivastava et al., 2014] of probability 0.5. Adam [Kingma and Ba, 2015] is used as an optimizer for all datasets with different step sizes determined by validation performances. ... = 0.05, λ = 10 4, and β = 0.9 in Cora, and is changed to 0.01 in Citeseer. ... = 1.0 and β = 0.5. We lastly set λ = 2 10 2 in Pub Med ... The number of diffusion operations is set to one in all datasets