Neural Network Playground
Neural Network Playground

Neural Network Playground

An interactive website for exploring neural networks and gradient descent through hands-on visualizations.

by GUAN Shui

Interactive experience Science & tech

Description

Machine Learning is an interactive educational website built with HTML, CSS, and JavaScript. It helps learners explore how neural networks make predictions and how gradient descent reduces prediction error. In the Neural Networks section, users can add input features, adjust the number of hidden neurons, and see the network architecture change. In the Gradient Descent section, users can adjust the learning rate, train the network step by step, and observe its progress through a three-dimensional loss landscape and a loss chart. A synthetic bike-demand dataset provides the training example. The project connects mathematical ideas with visual experiments, allowing users to explore how network structure and learning rate affect training. Its source code is hosted on Gitee, and updates are automatically deployed to Netlify through Gitee Go.

Desktop Homepage

Desktop Screenshot

Mobile Homepage

Mobile Screenshot

Open Graph Image

Open Graph Image

App Icon

App Icon

Favicon

Favicon