Keyphrases
Design Optimization
100%
Recurrent Neural Network
100%
Field Programmable Gate Arrays
100%
Recurrent Neural Network Model
57%
Long Short-term Memory
42%
Hardware Implementation
28%
FPGA Implementation
28%
Block-circulant
28%
Accuracy Requirements
28%
Block Size
28%
Irregular Networks
28%
Model Compression
28%
High Sensitivity
14%
Performance Improvement
14%
Energy Efficiency
14%
Simultaneous Model
14%
Network Structure
14%
Optimization Framework
14%
Implementation Model
14%
Time-accurate
14%
Activation Function
14%
Maximum Energy Efficiency
14%
Pruning
14%
Alternating Direction multiplier Method
14%
Challenging Tasks
14%
Neural Network Training
14%
Compression Ratio
14%
Weight Matrix
14%
Systematic Design
14%
Size Ratio
14%
Energy Efficiency Improvement
14%
Design Exploration
14%
Network Type
14%
Neural Network Implementation
14%
Matrix Representation
14%
Automatic Speech Recognition
14%
Neural Network Hardware
14%
Gated Recurrent Unit Neural Network
14%
Neural Network Framework
14%
Long Short-term Memory Neural Network
14%
Multiplier Technique
14%
Processing Element Design
14%
Weight Pruning
14%
Block Circulant Matrix
14%
Block Compression
14%
Model Acceleration
14%
Engineering
Design Optimization
100%
Recurrent Neural Network
100%
Network Model
33%
Long Short-Term Memory
33%
Field Programmable Gate Arrays
25%
Circulant
25%
Energy Conservation
16%
Block Size
16%
Energy Efficiency
16%
Recurrent
8%
Experimental Result
8%
Major Type
8%
Related Application
8%
Alternating Direction Method of Multipliers
8%
Compression Ratio
8%
Matrix Representation
8%
Automatic Speech Recognition
8%
Activation Function
8%
Neural Network Training
8%
Processing Element
8%
Computer Science
Recurrent Neural Network
100%
Design Optimization
100%
Neural Network Model
26%
Long Short-Term Memory Network
26%
Field Programmable Gate Arrays
20%
Hardware Implementation
13%
Accuracy Requirement
13%
Model Compression
13%
Energy Efficiency
13%
Speech Recognition
6%
Optimization Framework
6%
Network Structures
6%
Design Exploration
6%
Network Type
6%
Processing Element
6%
Matrix Representation
6%
Alternating Direction Method of Multipliers
6%
Neural Network Training
6%
Experimental Result
6%
Related Application
6%
Compression Ratio
6%
Activation Function
6%
Computer Hardware
6%
Gated Recurrent Unit Network
6%