Statistical Machine Learning (Team VivaLaVida-Lockdown)

This repository contains the source code for assignment 1 of the COMP90051 Statistical Machine Learning course at the University of Melbourne.

Submission Details

Team members:

• Matthias Bachfischer (Student ID: 1133751)

• Liam Simon (Student ID: 1128453)

• Parikshit Diwan (Student ID: 1110497)

Tean location: Melbourne

Project structure

• data/ -- Original dataset and preprocessed dataset in form of individual (source, target) pairs
• doc/ -- Documentation and report
• src/ -- Source code for the classifiers
• src/01_sklearn -- Source code for the SVM and ensemble classifiers
• src/02_fastgae -- Source code and configuration for the fastgae graph autoencoder. Parts of the source code were obtained from the fastgae repository
• src/03_biggraph -- Source code and configuration for the PyTorch biggraph classifier. Parts of the source code were obtained from the PyTorch BigGraph repository

System overview

This repository contains the source code to implement five classification systems for the task of predicting links between users in online social networks:

• A SVM classifier (Source)
• Two Ensemble classifiers (Source)
• A FastGAE Graph Autoencoder(Source) based on convolutional graph neural network (CGN)
• A prediction system built using the PyTorch BigGraph framework (Source).

For further information, please refer to the project report attached to this submission.