Sentiment Classification using RNN Variants

This project implements and compares multiple deep learning models for sentiment analysis on the IMDb movie reviews dataset.

Objective

To classify movie reviews as positive or negative using different recurrent neural network architectures and analyze their performance.

Dataset

• IMDb Movie Reviews Dataset
• Loaded using the HuggingFace datasets library
• Binary sentiment labels (positive / negative)

Models Implemented

• Simple RNN
• LSTM
• Bi-directional LSTM (BiLSTM)
• Multi-layer LSTM

Preprocessing Steps

• Tokenization using NLTK
• Stopword removal
• Vocabulary creation
• Sequence padding
• Train / test split

Technologies Used

• Python
• PyTorch
• HuggingFace Datasets
• NLTK
• NumPy

How to Run

1. Install dependencies:

pip install torch datasets nltk