Compression and Speed-up of Convolutional Neural Networks Through Dimensionality Reduction for Efficient Inference on Embedded Multiprocessor.
Lucas Fernández BrilletNicolas LeclaireStéphane ManciniMarina NicolasSébastien Cleyet-MerleJean-Paul HenriquesClaude DelnondedieuPublished in: J. Signal Process. Syst. (2022)
Keyphrases
- efficient inference
- convolutional neural networks
- dimensionality reduction
- probabilistic inference
- conditional random fields
- human pose estimation
- fully connected
- markov random field
- exact inference
- hidden variables
- structured prediction
- markov networks
- graph structure
- high dimensional
- low dimensional
- approximate inference
- principal component analysis
- factor graphs
- feature extraction
- high dimensional data
- graphical models
- linear models
- bayesian networks
- principal components
- first order logic
- support vector machine
- feature space
- pattern recognition
- pattern classification
- pose estimation
- higher order
- least squares
- data points
- probability distribution
- prior knowledge