NullHop: A Flexible Convolutional Neural Network Accelerator Based on Sparse Representations of Feature Maps.
Alessandro AimarHesham MostafaEnrico CalabreseAntonio Rios-NavarroRicardo Tapiador-MoralesIulia-Alexandra LunguMoritz B. MildeFederico CorradiAlejandro Linares-BarrancoShih-Chii LiuTobi DelbrückPublished in: CoRR (2017)
Keyphrases
- sparse representation
- feature maps
- convolutional neural network
- dictionary learning
- face detection
- sparse coding
- image classification
- self organizing maps
- face recognition
- neural network
- kernel function
- saliency map
- signal processing
- fuzzy neural network
- image representation
- high dimensional data
- feature vectors
- dimensionality reduction
- pattern recognition
- object detection
- computer vision
- maximum likelihood
- visual attention
- image data
- high dimensional
- feature space
- learning algorithm