The course is designed to give a wide perspective in machine learning, including but not limited to deep learning methods. The course will be online using notes on Jupyter Notebooks. The coding environment is Julia. The students are expected to install Julia and go over documentation on their own. The projects will also be coded in Julia and are expected to involve a comparative assessment of deep learning and at least one of other methods. Example codes in Julia on selected topics will be studied.

Curriculum (13 Weeks)

1. What is Machine Learning
2. Parametric Learning Basics – 2 Weeks
3. Regression Classics, eg. LS, Ridge, RLS – 1 Week
4. Classification Classics, eg. kNN, LR, LDA, SVM, RF, XGB – 2 Weeks
5. Projection onto Convex Sets, eg. POCS – 1 Week
6. Sparsity Promoting Learning, eg. LASSO – 1 Week
7. Kernel Methods, eg. GLM, k-SVM – 2 Weeks
8. Project Progress Presentations – 1 Week
9. Deep Learning – 3 Weeks
   1. The perceptron and feed-forward models
   2. Backpropagation
   3. Activation and loss functions
   4. Optimizers
   5. Main Architectures
      1. ​Linear Model & MLP
      2. CNN/LeNet & Batch Normalization
      3. RNN
      4. Variational Autoencoders
      5. GAN
10. Manifold Learning and Latent Variables (If time permits)