.png){kind=icon}
Return
A Memory-Efficient Learning Framework for Symbol Level Precoding With Quantized NN Weights
.png){kind=icon}
OA
AI
DOI：10.1109/OJCOMS.2023.3285790.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This paper proposes a memory-efficient deep neural network (DNN) framework-based sym-bol level precoding (SLP). We focus on a DNN with realistic finite precision weights and adopt an unsupervised deep learning (DL) based SLP model (SLP-DNet). We apply a stochastic quantization (SQ) technique to obtain its corresponding quantized version called SLP-SQDNet. The proposed scheme offers a scalable performance vs memory trade-off, by quantizing a scalable percentage of the DNN weights, and we explore binary and ternary quantizations. Our results show that while SLP-DNet pro-vides near-optimal performance, its quantized versions through SQ yield similar to 3.46x and similar to 2.64x model compression for binary-based and ternary-based SLP-SQDNets, respectively. We also find that our pro-posals offer similar to 20x and similar to 10x computational complexity reductions compared to SLP optimization-based and SLP-DNet, respectively.
Keywords:
Symbol-level-precoding
constructive interference
power minimization
deep neural networks (DNNs)
stochastic quantization (SQ)
Journal
I
IF:
4.3
Papers:
1.7K
Citations:
991
Organization
Cited Papers
.png){kind=icon}
ShareSaveA Hierarchical Cardiac Rhythm Classification Methodology Based on Electrocardiogram Fiducial Points
OAAI
OAAIShareSaveShareSaveShareSaveShareSave
View more