This GitHub project focuses on developing deep learning model for image classification using the MNIST dataset. The MNIST dataset consists of handwritten digits (0-9) and is commonly used as a benchmark for testing various machine learning algorithms.
The objective is to implement both a conventional Neural Network and a Convolutional Neural Network (CNN) for accurate digit recognition and comparison.
- Data Collection: The project utilizes the MNIST dataset, which is available through popular machine learning libraries like TensorFlow and PyTorch.
Model Architecture:
- Neural Network: A traditional feedforward neural network with fully connected layers for image classification.
- Convolutional Neural Network (CNN): Utilizes convolutional layers to automatically learn hierarchical features from the input images. Training and Evaluation: Implements training routines for both models and evaluates their performance using metrics like accuracy.
The following deep learning architectures have been implemented:
- Fully connected Neural Network: A standard feedforward neural network with multiple layers, including input, hidden, and output layers. This architecture learns to map input images to corresponding digit labels.
- Convolutional Neural Network (CNN): Leverages convolutional layers to capture spatial hierarchies in the input images. CNNs are particularly effective for image-related tasks, enabling automatic feature extraction.
Ashish Shivajirao Jadhav - @ashishjadhav