Octo: INT8 Training with Loss-aware Compensation and Backward Quantization for Tiny On-device Learning.
Qihua ZhouSong GuoZhihao QuJingcai GuoZhenda XuJiewei ZhangTao GuoBoyuan LuoJingren ZhouPublished in: USENIX Annual Technical Conference (2021)
Keyphrases
- supervised learning
- learning algorithm
- learning process
- learning problems
- reinforcement learning
- prior knowledge
- knowledge acquisition
- machine learning
- learning systems
- forward and backward
- learning tasks
- background knowledge
- online learning
- virtual environment
- serious games
- feedforward neural networks
- structured prediction
- decision trees
- recurrent networks
- motor skills
- online training