A NN/HMM hybrid for continuous speech recognition with a discriminant nonlinear feature extraction.
Gerhard RigollDaniel WillettPublished in: ICASSP (1998)
Keyphrases
- continuous speech recognition
- feature extraction
- discriminant analysis
- hidden markov models
- principal component analysis
- kernel principal component analysis
- neural network
- nearest neighbor
- appearance based face recognition
- image processing
- dimensionality reduction
- feature transformation
- kernel pca
- preprocessing
- pattern recognition
- feature vectors
- kernel fisher discriminant analysis
- face recognition
- handwritten word recognition
- support vector machine svm
- wavelet transform
- weight update
- feature set
- nonlinear functions
- knn
- independent components analysis
- feature space
- feed forward
- speech recognition
- hybrid models
- discriminant projection
- internal model control
- linear discriminant analysis
- artificial neural networks
- gabor filters
- neural network model
- dimension reduction
- hybrid intelligent systems
- frequency domain
- image classification
- genetic algorithm
- texture analysis
- feature selection
- high dimensional
- fuzzy logic
- face images
- low dimensional
- high dimensional data
- image feature extraction
- gesture recognition
- manifold learning
- pattern classification