CSCI-385 Machine Learning Projects
This repository contains three machine learning projects developed as part of the CSCI-385 Machine Learning course. The projects showcase the use of neural networks with PyTorch to solve different tasks using a Multilayer Perceptron (MLP) model for classification and prediction.
Projects Overview
-
GTSRB Dataset: Traffic Sign Classification
• Task: Classifying traffic signs from the German Traffic Sign Recognition Benchmark (GTSRB) dataset. • Model: Multilayer Perceptron (MLP). • Goal: Accurately identify traffic signs from a set of images, which is crucial for autonomous driving systems.
-
MNIST Dataset: Handwritten Digit Recognition
• Task: Classifying handwritten digits from the MNIST dataset. • Model: Multilayer Perceptron (MLP). • Goal: Achieve high accuracy in recognizing digits (0–9) from grayscale images.
-
Housing Price Dataset: Housing Price Prediction
• Task: Predicting house prices based on various features such as area, number of rooms, etc. • Model: Multilayer Perceptron (MLP). • Goal: Estimate housing prices accurately given the input features, which can be applied to real-world real estate analysis.
Features
• Datasets: All datasets were sourced from Kaggle, including the MNIST, GTSRB, and Housing Price data.
• Training and Evaluation: The models were trained and evaluated using PyTorch. Each model’s performance is displayed via loss function and accuracy function graphs.
• Neural Network Architecture: The MLP architecture varies depending on the dataset, tuned to optimize performance for each specific task.
How to Use
1. Clone the repository:
git clone
-
Install the required dependencies: pip install -r requirements.txt
Run each project by navigating to the respective folder and following the instructions in the README.md file within each folder.
Results
• Each project includes training logs, graphs for loss and accuracy, and the final trained model weights.