Group Assignment 1

Portfolio Exercise 1

Note: M3 - Group Assignment 1 Deadline: Wednesday 7th of February at 12.00 PM

Introduction

In this assignment, you are required to delve into the practical aspects of Deep Learning by constructing and evaluating a neural network using PyTorch. This exercise is designed to deepen your understanding of neural network architectures, hyperparameter tuning, and the preprocessing steps necessary for effective model training and evaluation. You will have the freedom to choose a dataset from either the M1 or M2 module or select an external dataset that intrigues you. By experimenting with different neural network configurations and hyperparameters, you will gain hands-on experience in optimizing ML models to achieve desired performance metrics.

Task

  • Build, train, and evaluate a neural network using Pytorch.
  • The neural network should have a minimum of 2 hidden layers.
  • Experiment with at least 5 different variations of hyperparameters (e.g., number of layers/neurons, activation functions, epochs, optimizers, learning rates, etc.).

The assignment should include the following steps:

  1. Feature Selection
  2. Feature Engineering (if necessary)
  3. Standard ML Preprocessing (if necessary)
  4. Train-Test Split
  5. Defining a Neural Network Architecture in Pytorch
  6. Defining a Training Loop
  7. Training the Model
  8. Experimenting with Different Hyperparameters
  9. Evaluating the Final Model on the Test Data

Data

  • Choose a dataset from the M1 or M2 module, or other datasets if you prefer.

Delivery

  • Create a GitHub repository.
  • Save the Colab notebook in the repository.
  • Provide a README.md with a brief description of the assignment.
  • Submissions can be made in groups of up to 3 members.
  • Submit the assignment by sending an email with the link to the repository to Hamid (hamidb@business.aau.dk).