jacklyonlee / ndr

Classical and Neural Dimensionality Reduction Techniques

Neural Dimensionality Reduction

This repository implements classical and neural dimensionality reduction techniques including Gaussian Random Projection, PCA, Autoencoder, Denoising Autoencoder, Variational Autoencoder and SimCLR and corresponding evaluation metrics such as Linear Probe, KNN and t-SNE.

Installation

• Set up and activate conda environment.

conda env create -f environment.yml
conda activate ndr

• Install pre-commit hooks.

pre-commit install

Quick Start

• Train and test a single model.

python train.py

• Run all experiments and plot results.

python run.py