DeepMTMV

• Deep Multimodality Model for Multi-task Multi-view Learning
• https://arxiv.org/abs/1901.08723

Summary:

• This project develops Deep-MTMV algorithm dealing with the multi-view multi-task classification problem.

Dataset

• Webkb dataset can be downloaded from the following link: http://www.cs.cmu.edu/afs/cs.cmu.edu/project/theo-20/www/data/
• Please extract webkb dataset in ./DeepMTMV-master/data/

To run the code:

• The code is written under python 3.6 with Pytorch framework and it could be ran by executing the following command:
• python DeepMTMV.py

Important Flags in the python scripts:

• -g: Specify the index of GPU to use if there are multiple GPUs. The default is 0;
• -d: Specify the directory of the Webkb dataset;
• -s: Specify the number of sample used during the training;

Note:

• The code requires the execution of Word2vec embedding, so Linux system or Mac OS is recommended.