Before the practice sessions
Please install Anaconda with Python version 3. You are allowed to use Python 2, but we cannot promise that everything we plan to do will be supported there. After installation make sure you are able to run jupyter notebook. Windows users open Anaconda prompt, other users just run from command line jupyter notebook. This should open a tab in your browser with your folder system. Go to a folder where you want your notebooks to be and create a new notebook by choosing New -> Python (from top right). After the notebook is open type in some commands like
import pandas as pd
import numpy as np
import sklearn
print("hello")
and press SHIFT + ENTER. If everything works then you are ready for the practice session. If some of the packages cannot be imported run conda install package-name to install the necessary packages.
If you have problems with this then write to the course forum in Piazza!
Practice Sessions
Practice 01 - Sep 9-13 - Basic linear classifier & Perceptron
Exercises, solutions, solutions as a Jupyter notebook
Practice 02 - Sep 16-20 - KNN & Naive Bayes
Exercises, Slides, solutions, solutions as a Jupyter notebook
Practice 03 - Sep 23-27 - Linear regression and regularization
Exercises (pdf), Exercises + solutions (colab)
Practice 04 - Sep 30-Oct 4 - Perceptron in dual form & SVM
Exercises (pdf), Solutions (colab)
Practice 05 - Oct 7-11 - Kernel methods
Oct 14 - TEST 1
Practice 06 - Oct 21-25 - Decision trees
Practice 07 - Oct 28-Nov 1 - F1 measure & ROC & Feature engineering
Practice 08 - Nov 4-8 - Logistic regression
Practice 09 - Nov 11-15 - Backpropagation & Softmax
Nov 18 - TEST 2
Practice 10 - Nov 25-29 - Ensemble methods
Practice 11 - Dec 2-6 - Probabilistic graphical models
Practice 12 - Dec 9-13 - Bayesian machine learning
Exercises, Notebook, Solutions 1 (pdf), Solutions 2 (notebook)