EvoSh: Evolutionary Search with Shaving to Enable Power-Latency Tradeoff in Deep Learning Computing on Embedded Systems.
Basar KütükçüSabur BaidyaAnand RaghunathanSujit DeyPublished in: SOCC (2023)
Keyphrases
- embedded systems
- deep learning
- evolutionary search
- low cost
- evolutionary algorithm
- unsupervised learning
- fitness function
- unsupervised feature learning
- weakly supervised
- machine learning
- software systems
- genetic algorithm
- search procedure
- power consumption
- feature selection
- field programmable gate array
- real time
- particle swarm optimization
- source code
- evolutionary computation
- search space
- search algorithm