A 1W8R 20T SRAM Codebook for 20% Energy Reduction in Mixed-Precision Deep-Learning Inference Processor System.
Ryotaro OharaMasaya KabutoMasakazu TaichiAtsushi FukunagaYuto YasudaRiku HamabeShintaro IzumiHiroshi KawaguchiPublished in: AICAS (2023)
Keyphrases
- deep learning
- unsupervised learning
- unsupervised feature learning
- random access memory
- energy consumption
- machine learning
- power consumption
- bag of words
- weakly supervised
- deep architectures
- feature vectors
- decision makers
- mental models
- decision making
- domain specific
- image classification
- maximum likelihood
- multiscale
- visual words
- low power
- bayesian networks
- image processing