ScaleCom: Scalable Sparsified Gradient Compression for Communication-Efficient Distributed Training.
Chia-Yu ChenJiamin NiSongtao LuXiaodong CuiPin-Yu ChenXiao SunNaigang WangSwagath VenkataramaniVijayalakshmi SrinivasanWei ZhangKailash GopalakrishnanPublished in: CoRR (2021)
Keyphrases
- lightweight
- communication overhead
- fully distributed
- distributed systems
- communication cost
- highly scalable
- image compression
- distributed control
- information sharing
- memory efficient
- computing environments
- scalable distributed
- multi agent
- spatially distributed
- distributed network
- distributed computation
- compression ratio
- computer networks
- online learning
- multiresolution
- cooperative
- multi party