Self-supervised Graph-level Representation Learning with Local and Global Structure

Authors: Minghao Xu, Hang Wang, Bingbing Ni, Hongyu Guo, Jian Tang

ICML 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments on both chemical and biological benchmark data sets demonstrate the effectiveness of the proposed approach.
Researcher Affiliation Academia 1Shanghai Jiao Tong University 2National Research Council Canada 3Mila Quebec AI Institute 4CIFAR AI Research Chair 5HEC Montr eal.
Pseudocode Yes Algorithm 1 Optimization Algorithm of Graph Lo G.
Open Source Code No The paper mentions checking 'released source code' for other methods but does not provide a link or statement for its own source code.
Open Datasets Yes In specific, a subset of ZINC15 database (Sterling & Irwin, 2015) with 2 million unlabeled molecules is employed for self-supervised pre-training. Eight binary classification data sets in Molecule Net (Wu et al., 2018) serve as downstream tasks...
Dataset Splits Yes Eight binary classification data sets in Molecule Net (Wu et al., 2018) serve as downstream tasks, where the scaffold split scheme (Chen et al., 2012a) is used for data set split.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory, or cloud instance types) used for running its experiments.
Software Dependencies No The paper mentions optimizers (Adam) and networks (GIN) but does not provide specific software dependencies with version numbers (e.g., Python, PyTorch, or other libraries with versions).
Experiment Setup Yes We use an Adam optimizer (Kingma & Ba, 2015) (learning rate: 1e-3) to pre-train the GNN... Unless otherwise specified, the batch size N is set as 512, and the hierarchical prototypes depth Lp is set as 3. For fine-tuning... an Adam optimizer (learning rate: 1e-3, fine-tuning batch size: 32) is employed to train the model for 100 epochs.