AdaBoost Project

This repository implements the AdaBoost algortihm in Julia and tests several AdaBoost-based machine learning models. The project includes scripts for training, testing, and analyzing models, as well as utilities for data processing and result visualization.

The most recent code and experiments can be found in the /adaboost directory, below an overview of its content.

Key Files

• tl_adaboost_mc.jl: Main implementation of AdaBoost for multi-class classification.
• tl_adaboost_mc_bin.jl: Binary classification version of AdaBoost.
• tl_adaboost_mc_bin_synth.jl: AdaBoost implementation for synthetic binary datasets.
• utility.jl: Utility functions for data processing, loss computation, and accuracy evaluation.
• setup.jl: Environment setup and configuration, when running on a cluster.
• analyze_weights.jl: Analysis of model weights during training.
• old_tl_adaboost_mc.jl: Old implementation of AdaBoost multi-class.

Folder Structure

• env_adaboostmc/: Julia environment files (Project.toml and Manifest.toml) for dependency management.
• logs/: Logs from model training and testing runs.
• results/: Results from experiments with multi-class AdaBoost.
• results_bin/: Results from binary classification experiments.
• results_bin_synth/: Results from synthetic binary dataset experiments.

Usage

Julia 1.9.3 was used for all the experiments.

1. Activate the Julia environment using the env_adaboostmc folder (run this command inside the /adaboost directory):

   julia --project=env_adaboostmc

2. In the Julia REPL run:

   using Pkg
   Pkg.instantiate()

   to install all the dependencies listed in the Manifest.toml file of the env_adaboostmc environment.