An Upper-Bound on the Required Size of a Neural Network Classifier

Keyphrases

• Neural Network 100%
• Neural Network Classifier 100%
• Training Data 60%
• Parameter-free 40%
• Training Set 20%
• Separability 20%
• Layer Width 20%
• Architectural Parameters 20%
• Classification Accuracy 20%
• Training Error 20%
• Activation Function 20%
• Training Examples 20%
• Data Dimensions 20%
• Rectified Linear Unit (ReLU) 20%
• Occam's Razor 20%

Computer Science

• Neural Network 100%
• Training Data 50%
• Free Parameter 33%
• Classification Accuracy 16%
• Training Example 16%
• Activation Function 16%
• Training Error 16%

Chemical Engineering

• Neural Network 100%