Deep-Neural-Network-Retrieval

This repo is initial implementation for the paper "Deep Neural Network Retrieval"

How to run

Take five-tasks (MNIST) as an example (containing three steps)

cd MNIST_five_tasks

First: preprocess dataset

python processdata.py

Second: train models

bash trainmodels.sh

Third:

Vector-based semantic feature extraction

python main_vector.py

Matrix-based semantic feature extraction

bash generatematrix.sh
python main_matrix.py