🤖 Unsupervised-learning

Note: This work summarizes my learnings in the domain of unsupervised learning as part of the "EECS 545: Machine Learning" course, conducted from January to April 2023. Here I provide implementations of topics such as Image compression, Face image dimension reduction, Audio separation, and Handwritten digits generation.

🎯 Goal

The goal of this project is to provide a useful resource for anyone seeking to understand and implement some of the fundamental unsupervised learning algorithms. For a brief overview, this repo contains the following implementations:

• K-means based image compression
• Gaussian Mixture Model with Expectation Maximization based image compression
• Principal component analysis for Eigenfaces generation
• Independent component analysis for Audio separation
• Conditional variational autoencoder based MNIST data generation

🛠️ Test/Demo

• Image compression
  • Launch the 'kmeans_gmm.ipynb 'jupyter notebook.
• Eigenfaces generation
  • Launch the 'pca.ipynb' jupyter notebook.
• Audio separation
  • Launch the 'ica.ipynb' jupyter notebook.
• MNIST data generation
  • Launch the 'cvae.ipynb' jupyter notebook.

📊 Results

📈 Image compression

• K-means

• Gaussian mixture model with EM

📈 Eigenfaces generation

📈 Audio separation

• Mixed audio Listen

• Unmixed source 0 Listen

• Unmixed source 1 Listen

• Unmixed source 2 Listen

• Unmixed source 3 Listen

• Unmixed source 4 Listen

📈 MNIST Data generation