Information-Maximized Soft Variable Discretization for Self-Supervised Image Representation Learning

This is a Pytorch implementation of paper.

Installation

Assuming Anaconda with python 3.8, a step-by-step example for installing this project is as follows:

conda install pytorch==1.12.1 torchvision==0.13.1 torchaudio==0.12.1 cudatoolkit=10.2 -c pytorch
conda install matplotlib

sbatch job_imsvd.sh  # need to adjust according to your computing platform

python linear_eval.py

Our pretrained model can be downloaded here

python visualize_matrix.py  # visualize the joint probability matrix

python visualize_samples.py  # visualize samples assigned to specific feature units

This project is under the CC-BY-NC 4.0 license. See LICENSE for details.

Name		Name	Last commit message	Last commit date
Latest commit History 3 Commits
LICENSE		LICENSE
README.md		README.md
augmentations.py		augmentations.py
eval_knn.py		eval_knn.py
imsvd.py		imsvd.py
job_imsvd.sh		job_imsvd.sh
linear_eval.py		linear_eval.py
resnet.py		resnet.py
visualize_matrix.py		visualize_matrix.py
visualize_samples.py		visualize_samples.py