TY - GEN
T1 - Low-Rank Decomposition for Rate-Adaptive Deep Joint Source-Channel Coding
AU - Xu, Man
AU - Lam, Chan Tong
AU - Liang, Yuanhui
AU - Ng, Benjamin
AU - Im, Sio Kei
N1 - Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - Deep joint source-channel coding (DJSCC) has received extensive attention in the communications community. However, the high computational costs and storage requirements prevent the DJSCC model from being effectively deployed on embedded systems and mobile devices. Recently, convolutional neural network (CNN) compression via low-rank decomposition has achieved remarkable performance. In this paper, we conduct a comparative study of low-rank decomposition for lowering the computational complexity and storage requirement for Rate-Adaptive DJSCC, including CANDECOMP/PARAFAC (CP) de-composition, Tucker (TK) decomposition, and Tensor-train (TT) decomposition. We evaluate the compression ratio, speedup ratio, and Peak Signal-to-Noise Ratio (PSNR) performance loss for the CP, TK, and TT decomposition with fine-tuning and pruning. From the experimental results, we found that compared with the TT decomposition, CP decomposition with fine-tuning lowers the PSNR performance degradation at the expense of higher compression and speedup ratio.
AB - Deep joint source-channel coding (DJSCC) has received extensive attention in the communications community. However, the high computational costs and storage requirements prevent the DJSCC model from being effectively deployed on embedded systems and mobile devices. Recently, convolutional neural network (CNN) compression via low-rank decomposition has achieved remarkable performance. In this paper, we conduct a comparative study of low-rank decomposition for lowering the computational complexity and storage requirement for Rate-Adaptive DJSCC, including CANDECOMP/PARAFAC (CP) de-composition, Tucker (TK) decomposition, and Tensor-train (TT) decomposition. We evaluate the compression ratio, speedup ratio, and Peak Signal-to-Noise Ratio (PSNR) performance loss for the CP, TK, and TT decomposition with fine-tuning and pruning. From the experimental results, we found that compared with the TT decomposition, CP decomposition with fine-tuning lowers the PSNR performance degradation at the expense of higher compression and speedup ratio.
KW - adaptive rate control image transmission
KW - deep joint source-channel coding
KW - low-rank decomposition
UR - http://www.scopus.com/inward/record.url?scp=85151696106&partnerID=8YFLogxK
U2 - 10.1109/ICCC56324.2022.10065853
DO - 10.1109/ICCC56324.2022.10065853
M3 - Conference contribution
AN - SCOPUS:85151696106
T3 - 2022 IEEE 8th International Conference on Computer and Communications, ICCC 2022
SP - 58
EP - 64
BT - 2022 IEEE 8th International Conference on Computer and Communications, ICCC 2022
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 8th IEEE International Conference on Computer and Communications, ICCC 2022
Y2 - 9 December 2022 through 12 December 2022
ER -