amannarsaria4/cs229

Solutions to the problem sets of CS229: Machine Learning from 2018

CS229: Machine Learning

This repository contains my solutions to the problem sets of the 2018 Stanford course CS229 by Andrew Ng.

I organized the solutions in IPython notebooks that can be read online in github. For better readability you can also view the notebooks in nbviewer.

Course material

• Syllabus of the 2018 version of the course
• Playlist of the lectures on YouTube
• Repository with the problem sets and many more resources for the course

Problem sets and solutions

• Problem set 0: Linear Algebra and Multivariable Calculus

  • Solutions Please note that I added a small aside on differentials and Jacobians, which I will consistently use throughout the problem sets though they haven't been covered in the lectures.

• Problem set 1: Supervised Learning

  • 1. Linear Classifiers (logistic regression and GDA)
  • 2. Incomplete, Positive-Only Labels
  • 3. Poisson Regression
  • 4. Convexity of Generalized Linear Models
  • 5. Locally weighted linear regression

• Problem set 2: Supervised Learning II

  • 1. Logistic Regression: Training stability
  • 2. Model Calibration
  • 3. Bayesian Interpretation of Regularization
  • 4. Constructing kernels
  • 5. Kernelizing the Perceptron
  • 6. Spam classification

• Problem set 3: Deep Learning & Unsupervised learning

  • 1. A Simple Neural Network
  • 2. KL Divergence and Maximum Likelihood
  • 3. KL Divergence, Fisher Information, and the Natural Gradient
  • 4. Semi-supervised EM
  • 5. K-means for compression

• Problem set 4: EM, DL, & RL

  • 1. Neural Networks: MNIST image classification
  • 2. Off Policy Evaluation And Causal Inference
  • 3. PCA
  • 4. Independent Components Analysis You can find the mixed and unmixed sound files here
  • 5. Markov Decision Processes