trsasasusu/graphkernelencodingallsubgraphsqc

Numerical experiments for Graph kernel encoding all subgraphs by superposition of quantum computing

Numerical experiments for Graph kernel encoding all subgraphs by superposition of quantum computing

Requirements

• python >=3.6
• pip3
• CUDA
• NVIDIA GPU with >=32510MiB memory

Installing

git clone https://github.com/TRSasasusu/GraphKernelEncodingAllSubgraphsQC
cd GraphKernelEncodingAllSubgraphsQC
pip3 install -r requirements.txt
cp grakel-datasets-base.py /usr/local/lib/python3.6/dist-packages/grakel/datasets/base.py

grakel-datasets-base.py fixes error when loading Fingerprint dataset by excluding graphs with #edges less than 1.

If needed, use virtualenv.

Usage

Computing kernel values

python3 calc_kernel.py <dataset> <kernel>

Compute kernel values for each pair of graphs.
For <dataset>, select from MUTAG, AIDS, ER_MD, PTC_FM, BZR_MD, IMDB-BINARY, Fingerprint and IMDB-MULTI.
For <kernel>, select from QK_BH_ve, QK_BH_ved, QK_SH_ve, QK_SH_ved, RW, GS and SP.

Performance evaluation

python3 evaluate.py <dataset> <kernel>

Evaluate performance of the kernel.
Make sure to run calc_kernel.py first.

The results are stored in performance_results.