This repository consists of the programming assignments that were implemented as a part of CS760 course. They are written in python and targeted to run on UWMadison Linux workstations.
Below are details of individual assignments:
-
Homework 2 - ID3 Decision Tree: Implementation of an ID3 Decision Tree learner for classification.
i. To characterize the predictive accuracy of the learned tree as a function of the training set size, Learning Curves were plotted and analyzed for two data sets - heart and diabetes. ii. To investigate how predictive accuracy varies as a function of tree size, plots of how test-set accuracy varies with various values of m (stopping criteria) were analyzed.
-
Homework 3 - Neural Network: A neural network with one hidden layer was trained with backpropagation to perform binary classification. n-fold stratified cross validation was used to train the network. The training was performed using Stochastic Gradient Descent, and a Sigmoid activation function was used.
i. To analyze the performance of the neural network (on sonar data set), the following graphs were plotted: Accuracy vs Epochs, Accuracy vs Number of folds, and ROC Curve
-
Homework 4 - Bayes Network: Implementation of Naive Bayes and Tree-Augmented Naive Bayes (TAN) was done to solve a binary classification problem on discrete valued features. Laplace estimates with pseudocounts of 1 were used to estimate the probabilities. Structure of Bayes network was determined using TAN algorithm.