Training strategies for deep latent models and applications to speech presence probability estimation

Mathematics

• Speech 100%
• Training 80%
• Mixture of Experts 75%
• Neural Networks 71%
• Strategy 53%
• Latent Variable Models 39%
• Latent Variables 34%
• Speech Enhancement 28%
• Model 24%
• Speech Processing 24%
• Back-propagation Algorithm 24%
• Network Architecture 23%
• Back Propagation 23%
• Noise Reduction 22%
• Expectation Maximization 20%
• Viability 18%
• Expectation-maximization Algorithm 18%
• Architecture 16%
• Experimental Results 15%
• Paradigm 14%
• Learning 13%
• Gradient 11%
• Context 9%
• Demonstrate 9%
• Standards 8%

Engineering & Materials Science

• Neural networks 46%
• Backpropagation 28%
• Speech processing 18%
• Speech enhancement 18%
• Noise abatement 15%
• Network architecture 14%