Deep Learning for Image and Text Analysis with Keras

Overview

This project demonstrates the application of deep learning techniques for analyzing both image and text datasets. It covers a comprehensive approach, including dataset preprocessing, model training, and evaluation. The project is structured into two main parts:

Analysis of the Olivetti Faces dataset using various machine learning and deep learning methods.

Sentiment analysis of the IMDB Movie Review Dataset employing models ranging from Naive Bayes to Recurrent Neural Networks (RNNs), including data augmentation techniques.

Datasets

• Olivetti Faces Dataset (AT&T Laboratories Cambridge): A collection of face images for testing facial recognition methods.
• IMDB Movie Review Dataset: A dataset for binary sentiment classification featuring a set of 25,000 highly polar movie reviews for training and 25,000 for testing.

Methodology

Image Analysis (Olivetti Faces)

• Data Preprocessing: Splitting the dataset into training and testing sets.
• Baseline Model: Implementing a simple classifier to set a baseline.
• Dimensionality Reduction and Classification: Using PCA and LDA for feature extraction, followed by SVM for classification.
• Deep Learning: Implementing a CNN based on the LeNet-5 architecture.

Text Analysis (IMDB Movie Reviews)

• Preprocessing: Cleaning and preparing text data.
• Baseline Model: Naive Bayes classifier for establishing a performance benchmark.
• RNN: Developing a Vanilla RNN model for sentiment analysis.
• Data Augmentation: Enhancing the dataset to improve model performance.

Running the Code

• Clone the repository to your local machine.
• Ensure all dependencies are installed.
• Open AIProject6.ipynb in Jupyter Notebook or Google Colab.
• Run the cells sequentially to reproduce the results.