1. School of Cybersecurity, Northwestern Polytechnical University, Xi’an 710072, China 2. Xingtang Communication Technology Limited Company, Beijing 100191, China 3. Aerospace DFH Satellite Limited Company, Beijing 100094, China
A novel semantic compression algorithm was proposed, aiming to address the inadequacies of traditional compression methods in retaining deep semantic information. The global correlated semantic importance (GCSI) was used as a semantic importance measurement parameter, and the semantic task relevance and semantic intrinsic relevance metrics were integrated to assess the importance of semantic features and achieve effective semantic compression. Experimental results show that the compression performance of the proposed algorithm is improved by more than 30% compared with traditional methods under different channel conditions, and the classification accuracy of the proposed algorithm is enhanced by more than 10% in low bandwidth and low signal-to-noise ratio (SNR) environments. Under the same bandwidth and performance requirements, the proposed algorithm exhibits superior noise stability compared to existing semantic compression methods based on semantic task relevance. The proposed algorithm significantly alleviates network transmission pressure, enhances task performance, and can meet the increasing data transmission requirements.
Yong LI,Zhiqiang LIU,Maoxing TIAN,Songlin JIA. Semantic compression algorithm based on global correlated semantic importance. Journal of ZheJiang University (Engineering Science), 2025, 59(4): 795-803.
Fig.2Network architecture of semantic compression algorithm based on global correlated semantic importance
Fig.3Additive white Gaussian noise channel model
算法
$\tau = 0.98 $
$\tau = 0.90 $
$\tau = 0.80 $
$\tau = 0.60 $
Acc
p
r
F1
Acc
p
r
F1
Acc
p
r
F1
Acc
p
r
F1
本研究
0.81
0.82
0.81
0.81
0.94
0.94
0.94
0.94
0.95
0.95
0.95
0.95
0.96
0.96
0.96
0.96
文献[25]、[28]
0.70
0.84
0.70
0.70
0.90
0.92
0.90
0.91
0.93
0.93
0.93
0.93
0.94
0.95
0.94
0.94
随机压缩
0.54
0.60
0.54
0.48
0.63
0.79
0.63
0.61
0.79
0.86
0.79
0.77
0.91
0.92
0.91
0.91
Tab.1Performance testing results of different compression algorithms in STL-10 dataset
Fig.4Classification accuracy of different compression algorithms at various signal-to-noise ratios (STL-10 dataset)
Fig.5Classification accuracy of different compression algorithms with different channel bandwidths
算法
$\tau = 0.98 $
$\tau = 0.90$
$\tau = 0.80 $
$\tau = 0.60 $
Acc
p
r
F1
Acc
p
r
F1
Acc
p
r
F1
Acc
p
r
F1
本研究
0.79
0.79
0.79
0.79
0.90
0.90
0.90
0.89
0.91
0.91
0.91
0.91
0.92
0.92
0.92
0.92
文献[25]、[28]
0.68
0.73
0.68
0.68
0.85
0.86
0.85
0.85
0.88
0.89
0.88
0.88
0.91
0.91
0.91
0.91
随机压缩
0.17
0.30
0.17
0.11
0.65
0.78
0.65
0.62
0.76
0.80
0.76
0.73
0.86
0.88
0.86
0.85
Tab.2Performance testing results of different compression algorithms in CIFAR-10 dataset
Fig.6Classification accuracy of different compression algorithms at various signal-to-noise ratios (CIFAR-10 dataset)
[1]
WEAVER W Recent contributions to the mathematical theory of communication[J]. ETC: A Review of General Semantics, 1953, 10 (4): 261- 281
[2]
SHANNON C E A mathematical theory of communication[J]. The Bell System Technical Journal, 1948, 27 (3): 379- 423
doi: 10.1002/j.1538-7305.1948.tb01338.x
[3]
张平, 戴金晟, 张育铭, 等 面向语义通信的非线性变换编码[J]. 通信学报, 2023, 44 (4): 1- 14 ZHANG Ping, DAI Jincheng, ZHANG Yuming, et al Nonlinear transform coding for semantic communications[J]. Journal on Communications, 2023, 44 (4): 1- 14
doi: 10.11959/j.issn.1000-436x.2023087
[4]
XIE H, QIN Z, LI G Y, et al Deep learning enabled semantic communication systems[J]. IEEE Transactions on Signal Processing, 2021, 69: 2663- 2675
doi: 10.1109/TSP.2021.3071210
[5]
TONG H, YANG Z, WANG S, et al. Federated learning based audio semantic communication over wireless networks [C]// Proceedings of the IEEE Global Communications Conference . Madrid: IEEE, 2021: 1–6.
[6]
HUANG D, TAO X, GAO F, et al. Deep learning-based image semantic coding for semantic communications [C]// Proceedings of the IEEE Global Communications Conference . Madrid: IEEE, 2021: 1–6.
[7]
PATWA N, AHUJA N, SOMAYAZULU S, et al. Semantic-preserving image compression [C]// Proceedings of the IEEE International Conference on Image Processing . Abu Dhabi: IEEE, 2020: 1281–1285.
[8]
SUN Q, GUO C, YANG Y, et al. Deep joint source-channel coding for wireless image transmission with semantic importance [C]// Proceedings of the IEEE 96th Vehicular Technology Conference . London: IEEE, 2022: 1–7.
[9]
WANG J, WANG S, DAI J, et al. Perceptual learned source-channel coding for high-fidelity image semantic transmission [C]// Proceedings of the GLOBECOM 2022-2022 IEEE Global Communications Conference . Rio de Janeiro: IEEE, 2022: 3959–3964.
[10]
WANG Q, SHEN L, SHI Y Recognition-driven compressed image generation using semantic-prior information[J]. IEEE Signal Processing Letters, 2020, 27: 1150- 1154
doi: 10.1109/LSP.2020.3004967
[11]
HU Q, ZHANG G, QIN Z, et al. Robust semantic communications against semantic noise [C]// Proceedings of the IEEE 96th Vehicular Technology Conference . London: IEEE, 2022: 1–6.
[12]
牛冠冲, 刘飞翔, 杨雯, 等 面向多任务的语义通信架构设计与实现[J]. 移动通信, 2024, 48 (2): 47- 55 NIU Guanchong, LIU Feixiang, YANG Wen, et al Design and implementation of semantic communication for goal-oriented multi-tasking[J]. Mobile Communications, 2024, 48 (2): 47- 55
doi: 10.3969/j.issn.1006-1010.20240109-0001
[13]
莫肇豪, 韦宝典, 马啸 基于上下文语义相似度的软压缩方法[J]. 移动通信, 2024, 48 (2): 90- 96 MO Zhaohao, WEI Baodian, MA Xiao Soft compression method based on context-semantic similarity[J]. Mobile Communications, 2024, 48 (2): 90- 96
doi: 10.3969/j.issn.1006-1010.20240111-0002
[14]
KUTAY E, YENER A. Classification-oriented semantic wireless communications [C]// 2024 IEEE International Conference on Acoustics, Speech and Signal Processing . Seoul: IEEE, 2024: 9096–9100.
[15]
GRASSUCCI E, PARK J, BARBAROSSA S, et al. Generative AI meets semantic communication: evolution and revolution of communication tasks [EB/OL]. (2024−01−10)[2024−07−15]. https://arxiv.org/pdf/2401.06803.
[16]
LI N, IOSIFIDIS A, ZHANG Q. Dynamic semantic compression for CNN inference in multi-access edge computing: a graph reinforcement learning-based autoencoder [EB/OL]. (2024−01−19)[2024−07−15]. https://arxiv.org/pdf/2401.12167.
[17]
WANG C, HAN Y, WANG W An end-to-end deep learning image compression framework based on semantic analysis[J]. Applied Sciences, 2019, 9 (17): 3580
doi: 10.3390/app9173580
[18]
SEBAI D. Multi-rate deep semantic image compression with quantized modulated autoencoder [C]// Proceedings of the IEEE 23rd International Workshop on Multimedia Signal Processing . Tampere: IEEE, 2021: 1–6.
[19]
DONG C, LIANG H, XU X, et al Semantic communication system based on semantic slice models propagation[J]. IEEE Journal on Selected Areas in Communications, 2023, 41 (1): 202- 213
doi: 10.1109/JSAC.2022.3221948
[20]
何晨光, 黄声显, 陈舒怡, 等 基于语义通信的低比特率图像语义编码方法[J]. 信号处理, 2023, 39 (3): 410- 418 HE Chenguang, HUANG Shengxian, CHEN Shuyi, et al A low bitrates image semantic coding method based on semantic communication[J]. Journal of Signal Processing, 2023, 39 (3): 410- 418
[21]
LUO S H, YANG Y Z, YIN Y L, et al. DeepSIC: deep semantic image compression [C]// Neural Information Processing . [S. l.]: Springer, 2018: 96–106.
[22]
PARK S, SIMEONE O, KANG J. End-to-end fast training of communication links without a channel model via online meta-learning [C]// Proceedings of the IEEE 21st International Workshop on Signal Processing Advances in Wireless Communications . Atlanta: IEEE, 2020: 1–5.
[23]
JIANG S, LIU Y, ZHANG Y, et al Reliable semantic communication system enabled by knowledge graph[J]. Entropy, 2022, 24 (6): 846
doi: 10.3390/e24060846
[24]
WANG Y, CHEN M, SAAD W, et al. Performance optimization for semantic communications: an attention-based learning approach [C]// Proceedings of the IEEE Global Communications Conference . Madrid: IEEE, 2021: 1–6.
[25]
刘传宏, 郭彩丽, 杨洋, 等 人工智能物联网中面向智能任务的语义通信方法[J]. 通信学报, 2021, 42 (11): 97- 108 LIU Chuanhong, GUO Caili, YANG Yang, et al Intelligent task-oriented semantic communication method in artificial intelligence of things[J]. Journal on Communications, 2021, 42 (11): 97- 108
doi: 10.11959/j.issn.1000-436x.2021214
[26]
ZITNICK C L, VEDANTAM R, PARIKH D Adopting abstract images for semantic scene understanding[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38 (4): 627- 638
doi: 10.1109/TPAMI.2014.2366143
[27]
KANG B Y. A novel approach to semantic indexing based on concept [C]// Proceedings of the 41st Annual Meeting on Association for Computational Linguistics . Sapporo: ACL, 2003: 44–49.
ZOU Bei-ji, GUO Jian-jing, ZHU Cheng-zhang, YANG Wen-Jun, WU Hui, HE Qi. Image classification based on BOW-HOG feature[J]. Journal of ZheJiang University (Engineering Science), 2017, 51(12): 2311-2319.
