Add Training Data Points

• Click anywhere on the canvas to add a training point.

• Use the radio buttons to select the class label:

  • 🔴 Class 1 (Red) – for one category
  • 🔵 Class 2 (Blue) – for the other category

• Add several points from both classes to build a dataset.

Start the Learning Simulation

• Click the "Start" button to initialize the perceptron with a random weight vector.
• The canvas will show the initial decision boundary (line) separating the two classes.

Step Through the Algorithm

• Click the "Step" button to perform one iteration of the perceptron update rule.
• Each step checks misclassified points and updates the weights.
• Observe the boundary updating with each step and how it improves classification.

Automate the Process

• Click "Automate" to continuously run the learning steps until convergence or manual stop.
• Use the Automation Speed slider to adjust how fast steps run (in milliseconds).

Adjust Learning Parameters

• Modify the Decay Rate using its slider. This influences how quickly the learning rate decays over time:

  • Low decay → stable learning
  • High decay → faster convergence but may underfit

• You can change this value before or during automation.

View Best Decision Boundary

• Click the "Show Best" button to toggle between:

  • The current boundary (latest weights)
  • The best boundary (lowest classification error so far)

Reset the Experiment

• Click "Reset" to clear:

  • All points on the canvas
  • Iteration count
  • Boundary and learning history

• Use this when starting a new dataset or experiment.

Monitor Progress

• View the iteration count just below the canvas.
• Watch the boundary shift and converge with each step or automated run.

Legend

• 🟢 Current Boundary – reflects the model’s most recent update
• 🟡 Best Boundary – represents the boundary with the fewest classification errors