GraphSage_Cross_Modal_GCN

Adaptive Semi-supervised Graph Convolutional Hashing Network For Large-Scale Cross-Modal Retrieval.

Overview

Here we provide the implementation of our model in TensorFlow, along with all experimental datasets. The repository is organised as follows:

• data/ contains the necessary dataset files. Here, we have three datasets including MIRFLICKR-25K, NUS-WIDE-10K and Wiki, which can be downloaded from pan.baidu.com: link: https://pan.baidu.com/s/1DlIxCvT_3vRKphMydnljVQ code: b6be.
• graphsage/ contains source code of GraphSage by Hanmiton et al. at https://github.com/williamleif/GraphSAGE.
• train_semi_flickr_adaptive.py execute a full training run on MIRFLICKR-25K dataset.
• train_semi_nuswide_adaptive.py execute a full training run on NUS-WIDE-10K dataset.
• train_semi_wiki_adaptive.py execute a full training run on Wiki dataset.
• evaluation.py contains evaluation metric and batch partition code.

Dependencies

The script has been tested running under Python 2.7, with the following packages installed (along with their dependencies):

• numpy==1.15.4
• tensorflow-gpu==1.12.0

In addition, CUDA 9.0 and cuDNN 7 have been used.