Accelerating Deep Learning Inference in Constrained Embedded Devices Using Hardware Loops and a Dot Product Unit.
Jure VrecaKarl J. X. SturmErnest GunglFarhad MerchantPaolo BientinesiRainer LeupersZmago BrezocnikPublished in: IEEE Access (2020)
Keyphrases
- embedded devices
- deep learning
- dot product
- embedded systems
- unsupervised learning
- mobile devices
- kernel function
- similarity function
- gaussian kernels
- feature space
- machine learning
- data storage
- mental models
- real time
- bayesian networks
- probabilistic inference
- limited memory
- security protocols
- low cost
- domain specific
- pattern recognition
- database systems
- mobile commerce
- learning algorithm