Subscribe via RSS
2024
Implementing Neural Network from scratch-Part 2 (Softmax Classification)
In this post, we implemented a neural network from scratch to perform multi-class classification on the MNIST dataset. We started by preprocessing the data, defining the network architecture, and implementing key components such as forward and backward propagation. By training the network, we minimized the error and improved its ability to classify handwritten digits accurately.
Implementing Neural Network from scratch-Part 1 (Binary Classification)
In this blog, we explored the process of building a neural network from scratch using Python and the MNIST dataset. By focusing on binary classification, we covered the essential components of neural networks, including data preprocessing, parameter initialization, forward pass, backpropagation, and training the network.
Implementing Stochastic Gradient Descent and variants from scratch.
In this post, we implemented stochastic gradient descent in python which is one of the efficient method for training ML models. The implementation encompases various SGD variants like constant and shrinking step sizes, momentum, and averaging, comparing how each one impacts the speed and accuracy of the model's convergence.