A neural network is a computational model inspired by the human brain's neural structure. It consists of interconnected nodes (neurons) that process and transmit information, enabling the network to learn patterns and make predictions.
Interactive Features
Add Points: Click on the canvas to add data points. First click adds class 0 (red), second click adds class 1 (blue).
Remove Points: Click on existing points to remove them.
Clear All: Remove all points and start fresh.
Network Types
XOR Problem: A classic problem demonstrating non-linear classification. The network learns to classify points based on the XOR logic gate pattern.
Circle Classification: The network learns to distinguish points inside and outside a circle, showcasing boundary learning.
Spiral Classification: A complex problem with interlocking spirals, testing the network's ability to learn intricate patterns.
Custom Points: Create your own classification problem by adding points manually.
Network Architecture
Learning Rate: Controls the speed of learning. Higher values (0.1) make learning faster but less stable, while lower values (0.001) make learning slower but more stable.
Hidden Layers: Additional layers between input and output. More layers can learn more complex patterns but require more training time.
Neurons per Layer: Number of processing units in each hidden layer. More neurons increase the network's capacity to learn complex patterns.
Visualization Guide
The colored background represents the network's decision boundary
Red points (class 0) and blue points (class 1) show your training data
Connected nodes display the network's structure and activation values
Line thickness and color indicate connection weights
Best Practices
Start with the XOR problem to understand basic classification
For complex patterns, increase neurons and layers gradually
If training is unstable, reduce the learning rate
Monitor loss and accuracy metrics to assess training progress
Experiment with different network architectures for each problem type