In this project, I implemented various neural network architectures to solve machine learning problems, including regression, image classification, and language identification.
Non-linear Regression: Designed and trained neural networks for modeling complex relationships in data.
Handwritten Digit Classification:Built a fully connected neural network to classify digits from the MNIST dataset, optimizing accuracy through hyperparameter tuning.
Language Identification: Developed a model to classify text samples by language using embedding techniques and PyTorch layers.
Convolutional Neural Networks (CNNs):Enhanced digit classification performance by implementing a CNN, leveraging convolutional layers for feature extraction.
This project required proficiency in PyTorch, hyperparameter tuning, batching for efficient computation, and systematic debugging to ensure robust architectures. By adhering to best practices, I optimised network performance and improved understanding of neural network design principles.