Click to place training points. Watch the Perceptron find the perfect line to separate classes!
Any linearly separable pattern can be learned - the future looks bright!
Click on the visualization to place training points
Training Points: 0
Epochs: 0
Convergence: Not Started
Optimism: Growing!
Phase 3: The XOR TrapIncomplete
The Impossible Problem
Can the mighty Perceptron learn Exclusive OR? Only when exactly ONE input is true...
Watch the machine struggle with what Minsky called "the fatal flaw."
Objective: Complete 3 training attempts to witness the mathematical impossibility.
Input A
Input B
XOR Output
0
0
0
0
1
1
1
0
1
1
1
0
Attempts: 0
Status: Ready to try
Hope: Fading...
Phase 4: Winter DescendsIncomplete
The Mathematical Proof
Minsky and Papert published "Perceptrons" in 1969. Watch the funding collapse unfold...
The first AI winter begins... but hope remains for deeper networks.
Research Funding: $1,000,000
Active Labs: 10
Published Papers: 5
Graduate Students: 50
"THINKING MACHINES - A DEAD END?" - The New York Times, 1970
But in hidden corners, researchers dream of deeper networks...
Level 4 Complete!
You've experienced the rise and fall of single-layer optimism.
Rosenblatt showed that artificial neurons could learn. Minsky showed their limits. The stage is set for the next breakthrough: hidden layers and the gradient as ghost.