Pith. sign in

REVIEW 3 major objections 2 minor 40 references

Spectral Efficiency-Aware Codebook Design for Task-Oriented Semantic Communications

T0 review · 3 major / 2 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Task-oriented semantic codebooks can be made both more accurate and more spectrally efficient by adding a Wasserstein-distance regularizer that pulls the codebook activation distribution toward the channel-optimal input distribution.

desk verdict The full manuscript text is a different paper, so the abstract's claims are unauditable—this needs to go back to the authors before any review. read the letter →

arxiv 2508.04223 v2 pith:CIKWFX6W submitted 2025-08-06 eess.SP eess.IV

Anbang Zhang , Shuaishuai Guo , Chenyuan Feng , Shuai Liu , Hongyang Du , Geyong Min This is my paper · ORCID
classification eess.SPeess.IV
keywords task-orientedsemanticcommunicationscodebookdesignspectralefficiencyWassersteindistanceactivationdistributionchannel-awarelatentrepresentationsdigitalcapacity-approaching
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tackles a hidden waste in task-oriented semantic communication: learned codebooks that map semantic features to channel symbols are often sparsely activated, so most symbols carry no useful task information. The authors build a codebook design framework that makes activation probability part of the optimization objective, so a codebook is judged by both inference accuracy and how much of the channel capacity it actually uses. They regularize the learned activation distribution toward the optimal channel input distribution with Wasserstein distance, the minimal transport cost between two distributions, and reinterpret that distance from a generative angle to fit semantic communication. Their proposed scheme, WS-DC, is reported in the abstract to outperform existing approaches in inference accuracy while significantly improving codebook efficiency. If that holds, semantic links can approach channel capacity without sacrificing the task they serve.

What carries the argument

The load-bearing component is a training objective in which the codebook activation probability enters explicitly, regularized by the Wasserstein distance between the learned activation distribution and the optimal channel input distribution. WS-DC (the Wasserstein-based adaptive hybrid distribution scheme) combines the task loss with this WS regularizer: the hybrid distribution is the codebook's activation distribution, and the Wasserstein term supplies channel-awareness. This machinery converts spectral efficiency from an incidental byproduct of quantization into an explicit, differentiable design target.

What would settle it

Train the same semantic encoder with and without the WS regularizer, logging activation frequencies and end-to-end inference accuracy across SNR from 0 to 20 dB. If every nonzero regularization weight makes accuracy worse than the unregularized baseline, or if the activation distribution's Wasserstein distance to a properly defined channel-optimal input distribution does not decrease when accuracy improves, the proposed mechanism is not what produces the reported gains.

Watch

Extended reading notes

Core claim

The paper claims that the sparse activation of learned codebooks is the main spectral-efficiency bottleneck in digital task-oriented semantic communication, and that this bottleneck can be optimized explicitly. The proposed WS-DC scheme learns compact, task-driven, channel-aware latent representations by adding a Wasserstein-distance term to the training loss; this term pushes the distribution of codebook activations toward the optimal channel input distribution, i.e., the input distribution that is best matched to the channel. The authors also reinterpret Wasserstein distance from a generative perspective, treating the codebook less as a fixed quantizer and more as a generator of channel in

Load-bearing premise

The claim stands or falls on whether pulling the codebook activation distribution toward the optimal channel input distribution can be done without erasing the differences between codewords that the inference task depends on.

Editorial extensions

If this is right

  • Task-oriented semantic systems using WS-DC should transmit fewer channel symbols for the same inference task, since rarely used codebook entries no longer consume capacity.
  • The channel-awareness added by the Wasserstein regularizer should make codebook performance transfer across signal-to-noise conditions, rather than being optimized for one fixed channel.
  • Reinterpreting Wasserstein distance in generative terms opens codebook design to generative-model tools, framing the codebook as a learned generator of near-optimal channel inputs.
  • The framework makes the tradeoff between inference accuracy and spectral efficiency explicit and tunable through the weight of the WS regularizer.
  • During deployment, the same activation-probability logging used in training gives an online diagnostic of how far the current codebook usage is from channel-optimal usage.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial note: the full-text body supplied under this title is a different manuscript, on hybrid quantum-HPC resource allocation, and contains no derivations or experimental detail for the abstract's claims; the extraction above therefore rests on the abstract alone.
  • The central compatibility assumption is testable: an explicit characterization of the optimal channel input distribution for each channel model would let one check whether WS-DC really approaches it, or whether the gains come from merely balancing codebook usage.
  • The reweighted objective resembles rate-distortion-perception tradeoffs, suggesting WS-DC could be viewed as a semantic analogue in which the perception constraint is replaced by a channel-capacity constraint on the codebook's output distribution.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 2 minor

Summary. The abstract proposes a spectral-efficiency-aware codebook design framework, WS-DC, for task-oriented semantic communications. The claimed method combines a task loss with a Wasserstein-distance regularizer that steers the learned codebook activation distribution toward an 'optimal channel input distribution', with experimental results said to show improved inference accuracy and codebook efficiency. However, the full text of the submitted manuscript is an entirely different paper on dynamic solutions for hybrid quantum-HPC resource allocation (arXiv:2508.04217). It contains no mention of semantic communications, codebooks, Wasserstein distance, the WS-DC scheme, or any experiment related to the abstract's claims. As submitted, the manuscript therefore provides no method description, no definitions, no theoretical development, and no empirical evidence for its central claims.

Significance. If the proposed WS-DC framework were fully developed and validated, it could be a useful contribution to codebook design for task-oriented semantic communications, particularly in addressing sparse codebook activation and spectral efficiency. However, because the submitted manuscript body is a different paper entirely, none of the claimed contributions can be evaluated. The significance of the work, assuming it exists, cannot be assessed from the submitted text.

major comments (3)
  1. [Abstract vs. Full Text] The central claims of the abstract are entirely unsupported by the manuscript body. The full text is 'Dynamic Solutions for Hybrid Quantum-HPC Resource Allocation' by Rocco et al., which discusses malleable job scheduling in HPC-QC systems and contains no reference to semantic communications, codebooks, Wasserstein distance, spectral efficiency, or WS-DC. Every load-bearing component of the claimed contribution—the architecture, the objective function, the definition of 'optimal channel input distribution', the generative reinterpretation of WS theory, and the experimental comparisons—is absent. As submitted, the paper cannot be reviewed for its stated content.
  2. [Abstract] Even taken on its own, the abstract does not define the 'optimal channel input distribution' that is used as the WS regularization target, nor does it explain how minimizing the Wasserstein distance to that target is compatible with preserving task-relevant information. The claim that WS-DC 'outperforms existing approaches in inference accuracy and significantly improves codebook efficiency' is presented without any of the necessary experimental details: datasets, baselines, channel models, communication protocols, metrics, or error bars. These omissions make the abstract's empirical claims unverifiable.
  3. [Full Text, Section I] The full text's introduction and subsequent sections describe a resource-allocation problem for hybrid quantum-HPC systems, including references to QPU scheduling, malleable jobs, and classical-quantum integration. None of this material connects to the abstract's task-oriented semantic communication framework. The manuscript therefore does not contain the proposed method or any of its supporting derivations, making the central claim of the paper unsupported by the submitted evidence.
minor comments (2)
  1. [General] The manuscript appears to contain the wrong full text relative to its abstract. At minimum, the title, author list, and abstract should correspond to the body. This is not a stylistic issue but a fundamental mismatch.
  2. [Abstract] The phrase 'optimal channel input distribution' is used without a formal definition or reference. If this distribution is meant to be capacity-achieving, that should be stated explicitly; if it is learned, its role in the WS regularizer needs clarification.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation can be found because the manuscript body is a different paper and contains none of the abstract's WS-DC framework, equations, or experiments.

full rationale

The abstract announces a spectral-efficiency-aware codebook design for task-oriented semantic communications with a Wasserstein-regularized objective (WS-DC). The supplied full text, however, is 'Dynamic Solutions for Hybrid Quantum-HPC Resource Allocation' by Rocco et al. (arXiv:2508.04217), and it contains no semantic-communication codebook model, no WS distance formulation, no definition of the 'optimal channel input distribution,' and no WS-DC experiments. There is therefore no derivation chain in the manuscript for the claimed results to reduce to. The abstract alone does not define the target distribution, so it cannot be shown that the target is equivalent to the learned codebook activation distribution by construction; the incompatibility of the WS regularizer with the task loss is asserted rather than derived. Unsupportedness or a mismatch between abstract and body is a completeness/correctness problem, not a circularity: no fitted parameter is renamed as a prediction, no self-citation carries a load-bearing uniqueness claim, and no equation is identical to another by construction. Accordingly, the circularity score is 0.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

Based only on the abstract, the identifiable free parameter is the regularization weight. The main axioms are domain assumptions about ToSC and the existence of an optimal channel input distribution, plus the paper-specific choice of Wasserstein distance. No new physical entities are introduced.

free parameters (1)
  • Wasserstein regularization weight
    The abstract mentions introducing WS distance as a regularization metric but does not state how its weight is chosen; it likely requires tuning. This is a free parameter affecting the balance between task performance and distribution matching.
assumptions (3)
  • domain assumption Task-oriented semantic communication only needs to transmit task-relevant information.
    This is the premise of ToSC as stated in the first sentence of the abstract.
  • domain assumption An optimal channel input distribution exists and is known or computable.
    The abstract refers to 'the optimal channel input distribution' without defining it; this is a necessary target for the WS regularization.
  • ad hoc to paper Wasserstein distance is an appropriate metric for measuring the gap between activation and input distributions.
    The paper chooses WS distance over other divergences; the abstract does not justify why WS is better, making it an ad hoc design choice.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Spectral Efficiency-Aware Codebook Design for Task-Oriented Semantic Communications." pith.science (2026). https://pith.science/paper/CIKWFX6W

@misc{pith2026250804223,
  author       = {Pith},
  title        = {Pith review of: Spectral Efficiency-Aware Codebook Design for Task-Oriented Semantic Communications},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CIKWFX6W}},
  note         = {Machine review of arXiv:2508.04223}
}
read the original abstract

Digital task-oriented semantic communication (ToSC) aims to transmit only task-relevant information, significantly reducing communication overhead. Existing ToSC methods typically rely on learned codebooks to encode semantic features and map them to constellation symbols. However, these codebooks are often sparsely activated, resulting in low spectral efficiency and underutilization of channel capacity. This highlights a key challenge: how to design a codebook that not only supports task-specific inference but also approaches the theoretical limits of channel capacity. To address this challenge, we construct a spectral efficiency-aware codebook design framework that explicitly incorporates the codebook activation probability into the optimization process. Beyond maximizing task performance, we introduce the Wasserstein (WS) distance as a regularization metric to minimize the gap between the learned activation distribution and the optimal channel input distribution. Furthermore, we reinterpret WS theory from a generative perspective to align with the semantic nature of ToSC. Combining the above two aspects, we propose a WS-based adaptive hybrid distribution scheme, termed WS-DC, which learns compact, task-driven and channel-aware latent representations. Experimental results demonstrate that WS-DC not only outperforms existing approaches in inference accuracy but also significantly improves codebook efficiency, offering a promising direction toward capacity-approaching semantic communication systems.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

40 extracted references · 40 canonical work pages

  1. [1]

    K. B. Letaief, Y. Shi, J. Lu, and J. Lu, ``Edge artificial intelligence for 6g: Vision, enabling technologies, and applications,'' IEEE J. Sel. Areas Commun., vol. 40, no. 1, pp. 5--36, Nov. 2022

    work page 2022

  2. [2]

    Y. Shi, K. Yang, T. Jiang, J. Zhang, and K. B. Letaief, ``Communication-efficient edge ai: Algorithms and systems,'' lEEE Commun. Surv. Tutorials, vol. 22, no. 4, pp. 2167--2191, July 2020

    work page 2020

  3. [3]

    W. Yang, Z. Q. Liew, W. Y. B. Lim, Z. Xiong, D. Niyato, X. Chi, X. Cao, and K. B. Letaief, ``Semantic communication meets edge intelligence,'' IEEE Wireless Commun., vol. 29, no. 5, pp. 28--35, Dec. 2022

    work page 2022

  4. [4]

    G. Liu, H. Du, D. Niyato, J. Kang, Z. Xiong, D. I. Kim, and X. Shen, ``Semantic communications for artificial intelligence generated content (aigc) toward effective content creation,'' IEEE Network, vol. 38, no. 5, pp. 295--303, Jan. 2024

    work page 2024

  5. [5]

    Q. Lan, D. Wen, Z. Zhang, Q. Zeng, X. Chen, P. Popovski, and K. Huang, ``What is semantic communication? a view on conveying meaning in the era of machine intelligence,'' J. Commun. Inf. Networks, vol. 6, no. 4, pp. 336--371, Dec. 2021

    work page 2021

  6. [6]

    S. Sun, Z. Qin, H. Xie, and X. Tao, ``Task-oriented scene graph-based semantic communications with adaptive channel coding,'' IEEE Trans. Wireless Commun., vol. 23, no. 11, pp. 17\,070--17\,083, Sept. 2024

    work page 2024

  7. [7]

    Y. Shi, Y. Zhou, D. Wen, Y. Wu, C. Jiang, and K. B. Letaief, ``Task-oriented communications for 6 G : Vision, principles, and technologies,'' IEEE Wireless Commun., vol. 30, no. 3, pp. 78--85, July 2023

    work page 2023

  8. [8]

    W. Xu, Z. Yang, D. W. K. Ng, M. Levorato, Y. C. Eldar, and M. Debbah, ``Edge learning for B 5 G networks with distributed signal processing: Semantic communication, edge computing, and wireless sensing,'' IEEE J. Sel. Topics Signal Process, vol. 17, no. 1, pp. 9--39, Jan. 2023

    work page 2023

Show all 40 references
  1. [9]

    L. Wang, W. Wu, F. Zhou, Z. Qin, and Q. Wu, ``Irs-enhanced secure semantic communication networks: Cross-layer and context-awared resource allocation,'' IEEE Trans. Wireless Commun., vol. 24, no. 1, pp. 494--508, Nov. 2025

    2025

  2. [10]

    Z. Qin, L. Liang, Z. Wang, S. Jin, X. Tao, W. Tong, and G. Y. Li, ``Ai empowered wireless communications: From bits to semantics,'' Proc. IEEE, vol. 112, no. 7, pp. 621--652, Aug. 2024

    2024

  3. [11]

    S. Guo, A. Zhang, Y. Wang, C. Feng, and T. Q. S. Quek, ``Semantic-enabled 6G communication: A task-oriented and privacy-preserving perspective,'' IEEE Network, pp. 1--1, Mar. 2025

    2025

  4. [12]

    H. Xie, Z. Qin, and G. Y. Li, ``Task-oriented multi-user semantic communications for VQA ,'' IEEE Wireless Commun. Lett., vol. 11, no. 3, pp. 553--557, Dec. 2022

    2022

  5. [13]

    Zhang, J

    H. Zhang, J. Ni, Z. Wu, X. Liu, and V. C. M. Leung, ``Resource allocation driven by large models in future semantic-aware networks,'' IEEE Wireless Commun., pp. 1--7, Mar. 2025

    2025

  6. [14]

    Z. Meng, C. She, G. Zhao, M. A. Imran, M. Dohler, Y. Li, and B. Vucetic, ``Task-oriented metaverse design in the 6 G era,'' IEEE Wireless Commun., vol. 31, no. 3, pp. 212--218, Feb. 2024

    2024

  7. [15]

    D. B. Kurka and D. Gündüz, ``Deepjscc-f: Deep joint source-channel coding of images with feedback,'' IEEE J. Sel. Areas Inf. Theory, vol. 1, no. 1, pp. 178--193, Apr. 2020

    2020

  8. [16]

    Weng and Z

    Z. Weng and Z. Qin, ``Semantic communication systems for speech transmission,'' IEEE J. Sel. Areas Commun., vol. 39, no. 8, pp. 2434--2444, June 2021

    2021

  9. [17]

    Jiang, C.-K

    P. Jiang, C.-K. Wen, S. Jin, and G. Y. Li, ``Deep source-channel coding for sentence semantic transmission with HARQ ,'' IEEE Trans. Commun., vol. 70, no. 8, pp. 5225--5240, June 2022

    2022

  10. [18]

    Y. Yang, C. Guo, F. Liu, L. Sun, C. Liu, and Q. Sun, ``Semantic communications with artificial intelligence tasks: Reducing bandwidth requirements and improving artificial intelligence task performance,'' IEEE Ind. Electron. Mag., vol. 17, no. 3, pp. 4--13, May 2023

    2023

  11. [19]

    F. Z. Safaeipour and M. Hashemi, ``Semantic-aware and goal-oriented communications for object detection in wireless end-to-end image transmission,'' in Int. Conf. Comput., Netw. Commun.,(ICNC), Big Island, HI, United states, Feb. 2024, pp. 182--187

    2024

  12. [20]

    Zhang, S

    H. Zhang, S. Shao, M. Tao, X. Bi, and K. B. Letaief, ``Deep learning-enabled semantic communication systems with task-unaware transmitter and dynamic data,'' IEEE J. Sel. Areas Commun., vol. 41, no. 1, pp. 170--185, Nov. 2023

    2023

  13. [21]

    D. Wen, P. Liu, G. Zhu, Y. Shi, J. Xu, Y. C. Eldar, and S. Cui, ``Task-oriented sensing, computation, and communication integration for multi-device edge AI ,'' IEEE Trans. Wireless Commun., vol. 23, no. 3, pp. 2486--2502, Aug. 2024

    2024

  14. [22]

    J. Shao, Y. Mao, and J. Zhang, ``Task-oriented communication for multidevice cooperative edge inference,'' IEEE Trans. Wireless Commun., vol. 22, no. 1, pp. 73--87, July 2023

    2023

  15. [23]

    Y. Wang, S. Guo, Y. Deng, H. Zhang, and Y. Fang, ``Privacy-preserving task-oriented semantic communications against model inversion attacks,'' IEEE Trans. Wireless Commun., vol. 23, no. 8, pp. 10\,150--10\,165, Mar. 2024

    2024

  16. [24]

    J. Shao, Y. Mao, and J. Zhang, ``Learning task-oriented communication for edge inference: An information bottleneck approach,'' IEEE J. Sel. Areas Commun., vol. 40, pp. 197--211, Nov. 2022

    2022

  17. [25]

    S. Xie, S. Ma, M. Ding, Y. Shi, M. Tang, and Y. Wu, ``Robust information bottleneck for task-oriented communication with digital modulation,'' IEEE J. Sel. Areas Commun., vol. 41, no. 8, pp. 2577--2591, June 2023

    2023

  18. [26]

    Q. Hu, G. Zhang, Z. Qin, Y. Cai, G. Yu, and G. Y. Li, ``Robust semantic communications with masked VQ-VAE enabled codebook,'' IEEE Trans. Wireless Commun., vol. 22, no. 12, pp. 8707--8722, Apr. 2023

    2023

  19. [27]

    S. Ma, W. Qiao, Y. Wu, H. Li, G. Shi, D. Gao, Y. Shi, S. Li, and N. Al-Dhahir, ``Task-oriented explainable semantic communications,'' IEEE Trans. Wireless Commun., vol. 22, no. 12, pp. 9248--9262, Apr. 2023

    2023

  20. [28]

    Esser, R

    P. Esser, R. Rombach, and B. Ommer, ``Taming transformers for high-resolution image synthesis,'' in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), Nashville, TN, USA, Nov. 2021, pp. 12\,868--12\,878

    2021

  21. [29]

    T. Kim, G. Song, S. Lee, S. Kim, Y. Seo, S. Lee, S. H. Kim, H. Lee, and K. Bae, ``L-verse: Bidirectional generation between image and text,'' in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), New Orleans, LA, USA, Sep. 2022, pp. 16\,505--16\,515

    2022

  22. [30]

    J. Yu, X. Li, J. Y. Koh, H. Zhang, R. Pang, J. Qin, A. Ku, Y. Xu, J. Baldridge, and Y. Wu, ``Vector-quantized image modeling with improved VQGAN ,'' in Proc. Int. Conf. Learn. Represent. (ICLR), Online, Jan. 2022

    2022

  23. [31]

    Zhang, Y

    A. Zhang, Y. Wang, and S. Guo, ``On the utility-informativeness-security trade-off in discrete task-oriented semantic communication,'' IEEE Commun. Lett., vol. 28, no. 6, pp. 1298--1302, Mar. 2024

    2024

  24. [32]

    Razavi, A

    A. Razavi, A. van den Oord, and O. Vinyals, ``Generating diverse high-fidelity images with VQ - VAE -2,'' in Proc. Adv. Neural Inf. Process. Syst. (NeurIPS), Vancouver, BC, Canada, vol. 32, Dec. 2019, pp. 14\,866--14\,876

    2019

  25. [33]

    Williams, S

    W. Williams, S. Ringer, J. Hughes, T. Ash, D. MacLeod, and J. Dougherty, ``Hierarchical quantized autoencoders,'' in Proc. Adv. Neural Inf. Process. Syst. (NeurIPS), Virtual, vol. 33, Dec. 2020, pp. 4524--4535

    2020

  26. [34]

    Takida, T

    Y. Takida, T. Shibuya, W. Liao, C.-H. Lai, J. Ohmura, T. Uesaka, N. Murata, S. Takahashi, T. Kumakura, and Y. Mitsufuji, `` SQ - VAE : Variational B ayes on discrete representation with self-annealed stochastic quantization,'' in Proc. Int. Conf. Mach. Learn. (ICML), Honolulu,...

    2022

  27. [35]

    Xie and Z

    H. Xie and Z. Qin, ``A lite distributed semantic communication system for internet of things,'' IEEE J. Sel. Areas Commun., vol. 39, no. 1, pp. 142--153, Nov. 2021

    2021

  28. [36]

    T.-Y. Tung, D. B. Kurka, M. Jankowski, and D. Gündüz, `` DeepJSCC-Q : Constellation constrained deep joint source-channel coding,'' IEEE J. Sel. Areas Inf. Theory, vol. 3, no. 4, pp. 720--731, Dec. 2022

    2022

  29. [37]

    W. Li, H. Liang, C. Dong, X. Xu, P. Zhang, and K. Liu, ``Non-orthogonal multiple access enhanced multi-user semantic communication,'' IEEE Trans. Cognit. Commun. Networking, vol. 9, no. 6, pp. 1438--1453, Aug. 2023

    2023

  30. [38]

    1em plus 0.5em minus 0.4em John Wiley & Sons, Ltd, Apr

    Entropy, Relative Entropy, and Mutual Information. 1em plus 0.5em minus 0.4em John Wiley & Sons, Ltd, Apr. 2005, ch. 2, pp. 13--55

    2005

  31. [39]

    K. Choi, K. Tatwawadi, A. Grover, T. Weissman, and S. Ermon, ``Neural joint source-channel coding,'' in Proc. Int. Conf. Mach. Learn. (ICML), Long Beach, California, USA, June 2019, pp. 1182--1192

    2019

  32. [40]

    Santambrogio, Optimal Transport for Applied Mathematicians: Calculus of Variations, PDEs, and Modeling

    F. Santambrogio, Optimal Transport for Applied Mathematicians: Calculus of Variations, PDEs, and Modeling. 1em plus 0.5em minus 0.4em Progress in Nonlinear Differential Equations and Their Applications, Oct. 2015, vol. 87

    2015

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.