An information maximization approach to overcomplete and recurrent representations

Keyphrases

• Information Maximization 100%
• Output Unit 100%
• Over-complete 100%
• Learning Rule 66%
• Recurrent Interactions 66%
• Two Dimensional 33%
• Mutual Information 33%
• Learning Algorithm 33%
• New Learning 33%
• Recurrent Connections 33%
• Limit of Zeros 33%
• Input Unit 33%
• Mutual Information Maximization 33%
• Feedforward Connection 33%

Engineering

• Maximization 100%
• Recurrent 100%
• Output Unit 60%
• Mutual Information 40%
• Learning Rule 40%
• Two Dimensional 20%
• Independent Component Analysis 20%
• Feedforward 20%
• Independent Component 20%
• Learning Algorithm 20%
• Input Unit 20%
• Dimensional Input 20%