Deep Hierarchical Graph Alignment Kernels

Authors: Shuhao Tang, Hao Tian, Xiaofeng Cao, Wei Ye

IJCAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Comparison with state-of-the-art graph kernels on various benchmark datasets demonstrates the effectiveness and efficiency of DHGAK. In this section, we analyze the performance of DHGAK compared to some state-of-the-art graph kernels. First, we give the experimental setup and introduce two realizations of DHGAK. Next, we demonstrate the performance of DHGAK on graph classification, the parameter sensitivity analysis of DHGAK, and the ablation study to reveal the contribution of each component in DHGAK. Finally, we compare the running time of each graph kernel.
Researcher Affiliation Academia 1Tongji University, Shanghai, China 2Jilin University, Changchun, China
Pseudocode Yes The pseudo-code is given in Algorithm 1 in B.1 1.
Open Source Code Yes The code is available at Github (https://github.com/EWesternRa/DHGAK).
Open Datasets Yes We evaluate DHGAK on 16 real-world datasets downloaded from [Kersting et al., 2016].
Dataset Splits Yes We use 10-fold cross-validation with a binary C-SVM [Chang and Lin, 2011] to test the classification performance of each graph kernel. The parameter C for each fold is independently tuned from {10 3, 10 2, . . . , 104} using the training data from that fold.
Hardware Specification Yes All experiments were conducted on a server equipped with a dual-core Intel(R) Xeon(R) Gold 6226R CPU @ 2.90GHz, 256 GB memory, and Ubuntu 18.04.6 LTS with 6 RTX 3090 GPUs.
Software Dependencies No The paper mentions software like BERT, word2vec, K-means, C-SVM, and Gra Ke L library, and the operating system Ubuntu 18.04.6 LTS, but does not provide specific version numbers for the main software components (e.g., ML frameworks, libraries) used in the experiments.
Experiment Setup Yes The parameters of DHGAK are set as follows. As discussed in Theorem 3, we set the number of clusters for different hops h as the same, around the size of the graph dataset, i.e., |Cψ| = |D| clusters factor, where the clusters factor are searched as base-10 log scale. We fix the experiment times T to 3 and other parameters are gridsearched as shown in Table 2 via 10-fold cross-validation on the training data. The parameter C for each fold is independently tuned from {10 3, 10 2, . . . , 104} using the training data from that fold. Table 2: parameter search ranges (BERT) fine-tune epochs {0, 3} b {0, 1, 2} H {1, 3, . . . , 9} α {0, 0.2, . . . , 1} clusters factor {0.1, 0.1 r, . . . , 0.1 r8, 2}