Skip to content

Official code for Interface Laplace Learning: Learnable Interface Term Helps Semi-Supervised Learning

Notifications You must be signed in to change notification settings

shwangtangjun/Inter-Laplace

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Interface Laplace Learning

This repository contains the code for Interface Laplace Learning: Learnable Interface Term Helps Semi-Supervised Learning

Dependencies

NumPy, SciPy and PyTorch. Scipy is for sparse matrix calculation. PyTorch is for acceleration of matrix multiplication on gpu. The code is tested on the following version:

numpy==1.26.4
scipy==1.11.4
torch==2.2.1

Files

data/ contains pretrained extracted features of each dataset. The files are collected from GraphLearning package without any change, but renamed for clarity.

inter_laplace.py includes the main function to perform training and inference.

utils.py includes utility functions.

preprocess.py includes the preprocess functions to get T, interface index and A.

Usage

The optimal parameters are provided in the appendix. Should take less than 1 second for each trial on gpu.

python inter_laplace.py --dataset mnist --label_num 1 --k_hop 4 --ridge 0.03

About

Official code for Interface Laplace Learning: Learnable Interface Term Helps Semi-Supervised Learning

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages