Auto-Tiler: Variable-Dimension Autoencoder with Tiling for Compressing Intermediate Feature Space of Deep Neural Networks for Internet of Things.
Jeongsoo ParkJungrae KimJong Hwan KoPublished in: Sensors (2021)
Keyphrases
- feature space
- neural network
- high dimension
- lower dimension
- feedforward neural networks
- pattern recognition
- mean shift
- high dimensional
- ubiquitous computing
- input space
- neural network model
- back propagation
- artificial neural networks
- mobile devices
- training set
- classification accuracy
- genetic algorithm
- low dimensional
- high dimensionality
- multi layer
- training process
- image retrieval
- image representation
- dimensionality reduction
- support vector machine
- key technologies
- physical world
- data sets
- feed forward
- dimension reduction
- feature selection
- self organizing maps
- input data
- feature vectors
- big data
- multilayer perceptron
- data compression
- kernel function
- location information
- mobile computing
- feature extraction
- data points