README for classification project

Link to see code and report: https://github.com/NicolaiAdil/ttt4275-classification_project

Introduction

Classification project in the course TTT4275 - Estimation, Detection and Classification. The project is divided into two parts, “The Iris task”, and “Classification of handwritten numbers 0-9” (MNIST). See the project description for more information.

Installation

Create a python virtual environment, where VENV_NAME is a variable name of your choice

python3 -m venv VENV_NAME

Source it (this is specific for Linux, if you have another OS look here)

source VENV_NAME/bin/activate

And install the necessary packages

pip3 install -r requirements.txt   

Running the Programs

When running the program a menu will show up. Choose the alternative you want to test.

Part 1: The Iris Task

python3 iris_task/main.py

Part 2: Classification of Handwritten Numbers (MNIST)

python3 mnist_task/main.py

Configuration

To properly test the efficiency of training in batches vs having everything in memory (Which is alternative 3 in the MNIST task), you need to change TEST_SAMPLE_SIZE to a larger number.