Skip to content

A model inversion technique that extracts ground truth labels before the reconstruction process and optimizes gradient using a cost function based on cosine similarity.

Notifications You must be signed in to change notification settings

wyfbw07/iDLG-with-cosine-similarity

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

iDLG with cosine similarity

This repo contains code for final project paper of ECSE 4964. The proposed method combines and involves two main components: Extracting ground truth labels before the reconstruction process and optimizing gradient using a cost function based on cosine similarity.

What's inside?

The script contains a sample reconstruction for three methods: The traditional DLG, improved DLG (iDLG), and iDLG with cosine similarity (our proposed method).

All three methods were configured identically using the LeNet model, the CIFAR-100 dataset, a learning rate of 0.1, 300 iterations, and the LBFGS optimizer.

How to use

Set up your Python environment and install the required packages. Then run the script.

Authors

Yixuan Zeng, Yifan Wang

Department of Electrical, Computer, and Systems Engineering

Rensselaer Polytechnic Institute

Troy, NY 12180

zengy5@rpi.edu, wangy77@rpi.edu

About

A model inversion technique that extracts ground truth labels before the reconstruction process and optimizes gradient using a cost function based on cosine similarity.

Resources

Stars

Watchers

Forks

Languages