Keyphrases
Aggregation Nodes
50%
Aggregation Operators
25%
Aggregator
100%
Associative Property
25%
Communication Cost
25%
Communication Time
25%
Compress
25%
Conventional Training
25%
Data Compression
25%
Deep Neural Network
100%
Distributed Systems
25%
Distributed Training
100%
Embedded Data
25%
Eon
25%
Floating-point
25%
Gradient Communication
25%
Gradient Update
25%
Hardware Efficiency
25%
Hardware-software Co-design
100%
In-network
75%
In-network Computing
50%
Lossy Compression Algorithms
25%
Multiple Streams
25%
Network Interface Card
50%
Network-centric
100%
Neural Network
50%
Order of Accuracy
25%
Precision Loss
25%
Process Training
50%
Training Algorithm
75%
Training Information
50%
Training System
25%
Unique Value
25%
Value Characteristics
25%
Computer Science
Characteristic Value
33%
Communication Cost
33%
Data Compression
33%
Deep Neural Network
100%
Direct Implementation
33%
Distributed Systems
33%
Embedding Data
33%
Floating Point
33%
Lossy Compression Algorithm
33%
Multiple Stream
33%
Network Interface Card
66%
Neural Network
66%
Participatory Design
100%
training algorithm
100%
Training Process
66%
Training System
33%