SGAE implementation in PyTorch

Project details

This repository contains a PyTorch implementation of the score-guided autoencdoer (SG-AE) presented in "Enhancing Unsupervised Anomaly Detection with Score-guided Network"

Requirements

• numpy==1.20
• pandas==1.1
• python==3.8.5
• pytorch==1.5.1
• sklearn==0.24
• tqdm==4.6

Documentation

• data: file path for datasets
• dataloader.py: definition of dataloader class, including data preprocess.
• sgae_main.py: entrance of experiments, including parameter adjustment.
• sgae_train.py: definition of training and testing.
• sgae.py: definition of model.

How to deploy

python sgae_main.py --[Parameter set] 

Links to datasets

• tabular data
  • attack
  • bcsc
  • Creditcard
  • diabetic
  • Donor
  • intrusion
  • market
• document
  • reuters/20news
• image
  • mnist