Pattern Recognition Weekly Coding Exercises

About This Repository

This repository contains my solutions to the weekly exercises of the 9th semester university subject "Pattern Recognition". Both reports and full problems statements are in Greek.

Week 1 - Random Variables and Optimization

In this exercise

• We calculate the probability mass function (PMF) of rolling a single die and then calculate the
  mean, variance, skewness, kurtosis of the random variable
• Create a simulation of throwing in dice and calculating experimentally all the above values using python
• Do the same for 2 dice
• Implement Gradient descent for known function
• Implement Newton Method for known function

Week 2 - Probability theory and Bayesian Decision Theory

In this weekly assignment we solve problems based on Bayesian Decision Theory.

Week 3 - Probabilistic Classifiers

In this exercise I create functions to

• Calculate decision boundary function for a known d dimension gaussian distribution with a known a priori
  probability
• Calculate Euclidian distance for d dimensions
• Calculate Mahalanobis distance

After that we use the above functions to solve problems using Maximum Likelihood Estimation

Week 4 - Non Parametric Estimation , Parzen Windows, K-NN

In this Exercise we create functions for Kernel Density Estimation Using Parzen Windows and KNN. These methods are then used to estimate the probability density function of a random variable

Week 5 - Linear Classifiers

In this exercise we implement the following algorithms:

• Batch perceptron
• Batch relaxation with margin
• MSE using Pseudoinverse
• Windrow-Hopf (LMS)
• Ho Kashyap
• Kesler Construction

And use them to classify the IRIS flower Dataset

Week 6 - Support Vector Machines

In this exercise we try different SVM for the IRIS flower Dataset using scikit-learn

Week 7 - Multi-Layered Perceptrons

Using scikit-learn we implement different MLP architectures for the IRIS Flower Dataset