Forward decoding kernel machines: A hybrid HMM/SVM approach to sequence recognition

Keyphrases

• Kernel Machines 100%
• Hidden Markov Model 100%
• Sequence Recognition 100%
• Maximum a Posteriori 40%
• Recursion 40%
• Probability Model 40%
• One-class Support Vector Machine (OCSVM) 40%
• State Transition Probability 40%
• Expectation Maximization 20%
• Probabilistic Model 20%
• State Transition 20%
• Posterior Probability 20%
• Conditional Probability 20%
• Phoneme Recognition 20%
• Transition Probability 20%
• Kernel-based 20%
• Nonzero Coefficients 20%
• Cross-entropy 20%
• Partially Labeled Data 20%
• Sequence Estimation 20%
• Recognition Accuracy 20%
• Recursive Scheme 20%
• Adaptive Sequencing 20%
• Noisy Labeled Data 20%
• Kernel Logistic Regression 20%
• Probability Regression 20%
• Classification Data 20%
• Speech Database 20%
• Sparse Kernel 20%
• TIMIT 20%
• SVM Method 20%
• Sum-product Algorithm 20%
• Kernel Extension 20%

Engineering

• State Transition 100%
• Maximum a Posteriori 66%
• Posterior Probability 33%
• Maximization 33%
• Recursive 33%
• Data Show 33%
• Observed Data 33%
• Product Algorithm 33%
• Nonzero Coefficient 33%
• Expectation Step 33%
• Recognition Accuracy 33%
• Conditional Probability 33%
• Support Vector Machine 33%

Computer Science

• Support Vector Machine 100%
• Kernel Machine 100%
• State Transition 60%
• Logistic Regression 20%
• Data Classification 20%
• Posterior Probability 20%
• Conditional Probability 20%
• Recognition Accuracy 20%
• Nonzero Coefficient 20%
• Product Algorithm 20%
• Kernel Expansion 20%