Pith. sign in

REVIEW 3 major objections 4 minor 1 cited by

From Raw Data to Structural Semantics: Trade-offs among Distortion, Rate, and Inference Accuracy

T0 review · 3 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read Sending a persistence diagram instead of the raw point cloud lets a receiver classify the object at roughly a 15-fold lower coded rate than an autoencoder-latent benchmark, and about 23-fold lower than raw data, even on a noisy binary…

desk verdict Novel idea—PDs as semantics—but the rate metric is misdefined and the benchmark is circular, so the headline gains don't hold. read the letter →

arxiv 2412.19825 v1 pith:K22JTNZU submitted 2024-12-18 cs.IT math.IT

classification cs.ITmath.IT
keywords persistencediagramssemanticcommunicationtopologicaldataanalysisrate-distortiontrade-offinferenceaccuracyquantizationBCHerror-correctingcodesstructuralsemantics
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

The paper argues that a persistence diagram—a multiset of birth–death pairs $(b,d)$ with $b

What carries the argument

The load-bearing object is the persistence diagram, a multiset of birth–death pairs $(b,d)$ in the open half-plane $b<d$, computed through a Vietoris–Rips filtration of the point cloud. The rate mechanism is a uniform 2D vector quantizer: each PD point is mapped to the center of one of $m^2$ cells in a bounded box, and the semantic rate is $R_{\mathrm{PD}} = M_S H(q_m)$ bits per object, where $M_S = m(m+1)/2$ counts the cells with nonzero probability and $H(q_m)$ is the entropy of the quantized symbol. The quantizer supplies both the distortion measure $D_{\mathrm{PD}}^{\mathrm{MSE}} = \mathbb{E}\|s - q_m(s)\|^2$ and the rate law used in all comparisons. On the inference side, PDs are converted to fixed-size vectors by a permutation-invariant neural layer so a standard classifier can be trained at the receiver; the same quantization pipeline is applied to autoencoder latents and raw points as baselines, and in noisy-channel experiments the quantized symbol stream is entropy-coded and protected by BCH error-correcting codes (cyclic codes that correct up to a prescribed number of bit errors).

What would settle it

Use the empirical distribution of $N_S$ from the paper's appendix to entropy-code each object's actual PD points and count the true per-object bits; if the coded PD rate is no longer roughly 15 times below the coded autoencoder-latent rate, or the 23-fold raw-data saving vanishes, the rate-efficiency claim is refuted.

Watch

Extended reading notes

Core claim

The central claim is that persistence diagrams, quantized and transmitted as symbols, preserve enough shape information for a downstream neural classifier to maintain high accuracy at rates where raw data and autoencoder latents fail. The paper shows this by computing PDs from Vietoris–Rips filtrations of the point clouds, quantizing each PD point with a uniform $m\times m$ grid on a bounded support box, and feeding the quantized coordinates through a permutation-invariant neural layer into a classifier. In a binary symmetric channel with crossover probabilities $\alpha=0.1$ and $\alpha=0.12$, BCH error-correcting codes wrapped around the quantized PD stream reach the error-free accuracy level at coded rates as low as 254.35 bits/object, about 15 times below the coded autoencoder-latent requirement; the same raw-data resource budget that infers one object supports about 23 PD-based inferences at comparable accuracy. The authors attribute this to sparsity of the PD distribution: probability mass concentrates in only $M_S$ of the $m^2$ cells, so semantic entropy is far lower than raw or latent coordinate entropy.

Load-bearing premise

The rate comparisons assume every object's persistence diagram can be described by exactly $M_S=m(m+1)/2$ quantized symbols, but the number of topological features per object is random, so the reported bits per object are not the actual source-coded message length.

Editorial extensions

If this is right

  • PD-based links can meet a target inference accuracy with far fewer bits than raw or learned-latent links, so the unused bit budget can be spent on forward error correction without exceeding the raw-data resource limit.
  • Because PD inference accuracy stays nearly flat as distortion and rate vary, an engineer can choose coarse quantization and still keep accuracy, simplifying the source code and improving tolerance to channel noise.
  • Under one raw-data transmission budget, a PD-based system can carry roughly 23 classifications at comparable accuracy, which translates directly into lower latency or higher throughput per unit bandwidth.
  • The rate-distortion gap between PD semantics and the alternatives is approximately a constant factor at the tested operating points, so the advantage is a structural property rather than a threshold artifact.

Reading between the lines

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

  • A testable extension is to recompute rates from each object's actual number of topological features: the rate formula counts $M_S=m(m+1)/2$ symbols per object, while $N_S$ per object is random (shown in the paper's own appendix), and real per-object message lengths may change the reported 15-fold and 23-fold comparisons.
  • The same quantization-plus-error-correction recipe should transfer to other structural summaries, such as witness complexes, cubical persistence, or persistent landscapes, whenever the summary distribution concentrates on few cells.
  • A natural next step is joint design of the quantization grid and the error-correcting code for a target accuracy, rather than fixing coarse quantization first and choosing a code afterwards.
  • The evidence opens a broader design question: when the transmitter's goal is a downstream task, the right semantic object may be the one with the sparsest distribution over the channel alphabet, not the one with the lowest raw-space distortion.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper proposes transmitting persistence diagrams (PDs) as 'structural semantics' from a sensor to a decision-maker, defines semantic distortion and rate for PDs through uniform 2D vector quantization, and empirically compares rate-distortion-accuracy trade-offs against raw point clouds and autoencoder (AE) latent representations. In a binary symmetric channel with BCH coding, it claims PDs achieve over 80% inference accuracy at a roughly 15 times lower rate than AE latents and a 23-fold resource reduction versus raw data. The central quantitative claim rests on the rate formula R_PD = M_S * H(q_m) in Eq. (13), where M_S is the number of nonzero-probability quantization regions rather than the number of transmitted PD points per object. The paper also provides an appendix treating the number of PD points as random, which exposes an internal inconsistency in the rate definition.

Significance. The idea of using topological summaries as communication semantics is novel and timely, and the paper contributes a complete experimental pipeline: VR-filtration PDs, PersLay-based classifiers, cross-validation, and publicly available code/data. The empirical observation that PD point distributions are highly sparse is interesting and could motivate efficient coding schemes. However, the headline rate-reduction numbers are built on a rate definition that does not count the actual message length per object, so the quantitative contributions are not currently supported. The conceptual framework is potentially salvageable after a corrected rate formulation and re-evaluated experiments, but the order-of-magnitude efficiency claims must be substantially revised.

major comments (3)
  1. [§V-B, Eq. (13); Appendix A and Fig. 8a] The semantic rate R_PD = M_S * H(q_m) does not correspond to the number of bits required to transmit a PD. A PD is a multiset of N_S points, each mapped to one alphabet symbol by the quantizer, so the per-object message length is the sum of N_S symbols plus the cost of conveying N_S (or a termination mechanism). The factor M_S = m(m+1)/2 is the number of quantization cells with nonzero probability, not the number of transmitted points. The paper itself models N_S as a random variable in Appendix A, and Fig. 8a shows a broad distribution with values up to about 150, so the rate definition is internally inconsistent. The same defect appears in §V-C and §V-D, where M_A = m^2 and M_G = m^2 are used instead of the actual numbers of transmitted points (d for AE, N_G for raw data). Consequently, the values r_PD = 30.58, r_AE = 466.19, and r_Raw = 6035.20 bits/object reported in §VI-C4 are not actual message lengths.
  2. [§VI-C4, Eq. (16) and Fig. 7] Because the 15-fold and 23-fold rate reductions are computed by applying BCH overhead to the incorrect r_PD and r_AE values, the paper's central efficiency claims are not established. With a correct rate computation, the AE rate for d=27 and m=10 would be d * H(q_a) = 27 * H(q_a) rather than 100 * H(q_a), while the PD rate would be E[N_S] * H(q_s) rather than 55 * H(q_s). Depending on E[N_S], the reported 15x advantage over AE could shrink by roughly an order of magnitude. The authors should re-derive the rates and rerun the comparisons before making any efficiency claim.
  3. [§VI-A, class label definition] The three classes are defined by the number of loops in the handwritten digit (C1 = one loop, C2 = two loops, C3 = no loops), which is exactly the topological feature captured by H1 of the VR filtration. This makes the classification task intrinsically aligned with the PD representation and creates a favorable comparison against raw data and AE latent representations, whose features are not constructed from this prior knowledge. To support a general claim that PD semantics improve communication efficiency, the evaluation should include tasks whose labels do not coincide so directly with the transmitted topology, or the paper should at least clearly acknowledge this bias.
minor comments (4)
  1. [Fig. 7 caption] There is a typo in the caption: 'PD semnatics' should be 'PD semantics'.
  2. [§VI-B, Eq. (14)] In Eq. (14), A_PD(m,t) is described as 'average error' but is then used as inference accuracy; please fix the terminology.
  3. [§V-A, §V-B] The paper repeatedly describes the distortion and rate definitions as 'qualitative', but Eqs. (8) and (13) are quantitative definitions; this wording should be adjusted.
  4. [Fig. 3 and Fig. 6] No confidence intervals or variability bands are reported for the estimated rates and distortions in Figs. 3 and 6; given the small dataset (600 images), bootstrap or repeated subsampling would strengthen the evidence.

Circularity Check

1 steps flagged · score 6.0 of 10

PD rate Eq. (13) is defined as the support-size M_S times the per-symbol entropy H(q_m) rather than the actual per-object symbol count N_S, so the advertised 15x/23x rate reductions are built into the definition.

  1. self definitional [Section V-B, Eq. (13); applied in Section VI-C4 (r_PD = 30.58 and the 15x/23x claims)]
    "Given PD semantics S = {s1, . . . ,sNS }, ... Qm(S) = {qm(s1), . . . , qm(sNS )}, (6) ... Finally, we define the rate of PD semantics as RPD [bits/object], where RPD = MS H(qm). Note that MS = m(m + 1)/2 and is introduced since pk is nonzero only for MS quantization regions, cf. Figure 2."

    In Eq. (6), each PD point s_n is quantized to exactly one alphabet symbol, so the source message for one object is a sequence of N_S symbols, where N_S is the random number of topological features. Figure 8a shows that N_S is a widely spread random quantity, unrelated to the fixed grid-geometry constant M_S = m(m+1)/2. Eq. (13) nevertheless multiplies the per-symbol entropy H(q_m) by M_S and labels the result 'bits/object.' The same template is applied to AE-latent representations with M_A = m^2 and to raw data with M_G = m^2, so the PD-vs-AE rate gap inherits a built-in geometric factor M_S/M_A ≈ 1/2 and, more importantly, ignores the actual number of transmitted points.

full rationale

The paper's simulation pipeline is otherwise self-contained: PDs are computed with GUDHI, vectorized by PersLay, classifiers are trained with cross-validation, and the BCH/BSC behavior is simulated with actual codes. Self-citations (e.g., [8], [11], [15], [47], [48]) appear only as background or related work and do not carry the argument. The one load-bearing circular element is the definition of semantic rate. Eq. (6) makes explicit that each PD point becomes one alphabet symbol, so an object's message is N_S symbols long; Eq. (13) instead sets R_PD = M_S * H(q_m) with M_S = m(m+1)/2, a geometric constant unrelated to the random N_S. The analogous definitions for AE and raw data use M_A = M_G = m^2. Consequently, the reported bits/object values (r_PD = 30.58, r_AE = 466.19, r_Raw = 6035.20) are not source-coded message lengths, and the celebrated 15x/23x rate reductions are partially baked into the chosen normalization rather than measured from actual Huffman/BCH streams. This makes the central rate comparison partially circular, although the sparsity of the estimated PD distribution, the distortion-accuracy robustness, and the BCH error-correction results remain independent empirical observations.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central quantitative claims rest on two free parameters (grayscale threshold, VR bound) and, critically, on the paper's own rate definition in Eq. (13). The rate definition implicitly assumes each object transmits M_S symbols, which is not the case when N_S varies. The benchmark task is also constructed from the topological invariant the representation encodes.

free parameters (2)
  • grayscale threshold (0.70) = 0.70
    Hand-chosen binarization threshold for extracting point clouds from MNIST images (Section VI-A). It determines which pixels become points, and hence shapes all downstream PDs and results.
  • VR filtration bound gamma_t = 16
    Hand-chosen maximum scale for the Vietoris-Rips filtration (Appendix B). Determines when all topological features appear and thus the content of each persistence diagram.
assumptions (4)
  • domain assumption PD points for an object are IID draws from a fixed compactly supported density fs.
    Assumption 1 in Section IV-A. The entropy and distortion formulas treat topological features as exchangeable IID samples; in reality H0 and H1 features have different birth-death statistics and are correlated through the filtration.
  • ad hoc to paper The semantic rate equals M_S times the per-symbol entropy of the quantizer.
    Eq. (13) in Section V-B defines R_PD = M_S * H(q_m) without derivation. This is the paper's own postulate, not a rate derived from an actual source code for the PD point sequence.
  • ad hoc to paper Persistence diagrams are sufficient semantics for the downstream classification task.
    The task labels are defined by loop counts, which are exactly the H1 persistent-homology invariants encoded in the PD (Section VI-A).
  • standard math PersLay vectorizations of PDs can be learned with a permutation-invariant neural network layer.
    Relies on the PersLay construction [19]; used as classifier input in Section III and Appendix B.

how reviews work

0 comments
Cite this review

Pith. "Pith review of From Raw Data to Structural Semantics: Trade-offs among Distortion, Rate, and Inference Accuracy." pith.science (2026). https://pith.science/paper/K22JTNZU

@misc{pith2026241219825,
  author       = {Pith},
  title        = {Pith review of: From Raw Data to Structural Semantics: Trade-offs among Distortion, Rate, and Inference Accuracy},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/K22JTNZU}},
  note         = {Machine review of arXiv:2412.19825}
}
read the original abstract

This work explores the advantages of using persistence diagrams (PDs), topological signatures of raw point cloud data, in a point-to-point communication setting. PD is a structural semantics in the sense that it carries information about the shape and structure of the data. Instead of transmitting raw data, the transmitter communicates its PD semantics, and the receiver carries out inference using the received semantics. We propose novel qualitative definitions for distortion and rate of PD semantics while quantitatively characterizing the trade-offs among the distortion, rate, and inference accuracy. Simulations demonstrate that unlike raw data or autoencoder (AE)-based latent representations, PD semantics leads to more effective use of transmission channels, enhanced degrees of freedom for incorporating error detection/correction capabilities, and improved robustness to channel imperfections. For instance, in a binary symmetric channel with nonzero crossover probability settings, the minimum rate required for Bose, Chaudhuri, and Hocquenghem (BCH)-coded PD semantics to achieve an inference accuracy over 80% is approximately 15 times lower than the rate required for the coded AE-latent representations. Moreover, results suggest that the gains of PD semantics are even more pronounced when compared with the rate requirements of raw data.

Figures

Figures reproduced from arXiv: 2412.19825 by the authors.

Figure 1
Figure 1. System model highlighting different stages. [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Uniform 2D vector quantization. of Si are vectors in the nonnegative orthant R 2 +, we model the semantic input to the quantizer as a sequence of 2D analog random variables. In particular, the input to the quantizer will be {Si}i∈N, where Si = {si,1, . . . , si,NSi } is the PD semantics of raw data Gi , si,j ∈ R 2 , and NSi is a finite integer specific to Si . Moreover, the elements of Si for all i are assumed to be… view at source ↗
Figure 1
Figure 1. Note that u and uˆ are not necessarily identical unless the communication between TX and RX is perfect. V. PD-SEMANTIC DISTORTION AND RATE In this section, we quantify the distortion and rate of PD semantics. The purpose is to use them as a basis for subsequent empirical evaluations of the trade-offs, cf. § III. Finally, analogous definitions for AE and raw data are also used in the empirical evaluations as a benchm… view at source ↗
Figures from the paper (6 more)
Figure 2
Figure 2. Figure 2: This [PITH_FULL_IMAGE:figures/full_fig_p006_2.png]
Figure 3
Figure 3. Figure 3: Empirical trade-off between DˆMSE and Rˆ. DPD MSE [cf. (8)] and RPD [cf. (11), (12), and (13)], we use an estimate ˆfs of fs based on the re-scaled histogram of 2D analog vectors si,1, . . . , si,NSi that constitute Si for all i ∈ N test t . More specifically, B is par…
Figure 4
Figure 4. Figure 4: Empirical probability distributions ˆfs [(a) and (d)], ˆf 27 a [(b) and (e)], and ˆfg [(c) and (f)] for different m. curves are obtained by changing the bins per dimension m from 10 to 27 [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: shows the empirical trade-off curves between infer￾ence accuracy and MSE distortion. The thick curves depict the average results that correspond to trained NN models for classification as discussed in § VI-B [cf. (14)] and the shaded regions highlight their variability…
Figure 6
Figure 6. Figure 6: Empirical trade-off between Aˆ and Rˆ. to PD semantics with rPD [bits/object], the raw data needs to operate with m = 27 which corresponds to a rate of rRaw = 6035.20 [bits/object] on average. These rate limits are depicted by using dotted vertical lines in [PITH_FULL…
Figure 8
Figure 8. Figure 8: Empirical probability distributions where [PITH_FULL_IMAGE:figures/full_fig_p012_8.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Resilient-native and Intelligent NextG Systems

    cs.ET 2025-06 unverdicted novelty 3.0 of 10

    A position paper defining resilience for 6G networks as resisting, recovering from, and adapting to unforeseen disruptions, with a proposed multi-disciplinary mathematical toolkit.

Reference graph

Works this paper leans on

65 extracted references · 41 canonical work pages · cited by 1 Pith paper

  1. [1]

    Shannon and W

    C. Shannon and W. Weaver, The Mathematical Theory of Communica- tion, ser. Illini books. University of Illinois Press, 1949

  2. [2]

    A mathematical theory of communication,

    C. E. Shannon, “A mathematical theory of communication,” The Bell system technical journal , vol. 27, no. 3, pp. 379–423, 1948

  3. [3]

    An outline of a theory of semantic information,

    R. Carnap and Y . Bar-Hillel, “An outline of a theory of semantic information,” Research Laboratory of Electronics, MIT , 1952

  4. [4]

    An outline of a theory of semantic information,

    ——, “An outline of a theory of semantic information,” Journal of Symbolic Logic, vol. 19, no. 3, pp. 230–232, 1954

  5. [5]

    Semantic information,

    Y . Bar-Hillel and R. Carnap, “Semantic information,” The British Journal for the Philosophy of Science , vol. 4, no. 14, pp. 147–157, 1953

  6. [6]

    What is semantic communication? A view on conveying meaning in the era of machine intelligence,

    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,” Journal of Communications and Information Networks , vol. 6, no. 4, pp. 336–371, 2021

  7. [7]

    6G networks: Beyond Shannon towards semantic and goal-oriented communications,

    E. C. Strinati and S. Barbarossa, “6G networks: Beyond Shannon towards semantic and goal-oriented communications,” Computer Net- works, vol. 190, p. 107930, 2021

  8. [8]

    Semantics-native com- munication via contextual reasoning,

    H. Seo, J. Park, M. Bennis, and M. Debbah, “Semantics-native com- munication via contextual reasoning,” IEEE Transactions on Cognitive Communications and Networking , vol. 9, no. 3, pp. 604–617, 2023. xii 0 50 100 150Topologicalfeatures(NS) 0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16Probability (a) (b) (c) 20 40 60 80 1001201401601802002-dimensionalpoints(NG...

Show all 65 references
  1. [9]

    Semantic communications for future internet: Fundamentals, applications, and challenges,

    W. Yang, H. Du, Z. Q. Liew, W. Y . B. Lim, Z. Xiong, D. Niyato, X. Chi, X. Shen, and C. Miao, “Semantic communications for future internet: Fundamentals, applications, and challenges,” IEEE Communications Surveys & Tutorials, vol. 25, no. 1, pp. 213–250, 2022

  2. [10]

    Semantic communications: Principles and challenges,

    Z. Qin, X. Tao, J. Lu, W. Tong, and G. Y . Li, “Semantic communications: Principles and challenges,” arXiv preprint arXiv:2201.01389 , 2021

  3. [11]

    Goal-oriented communications for the IoT and application to data compression,

    C. Zhang, H. Zou, S. Lasaulce, W. Saad, M. Kountouris, and M. Bennis, “Goal-oriented communications for the IoT and application to data compression,” IEEE Internet of Things Magazine , vol. 5, no. 4, pp. 58– 63, 2022

  4. [12]

    Semantics-empowered communications: A tutorial-cum-survey,

    Z. Lu, R. Li, K. Lu, X. Chen, E. Hossain, Z. Zhao, and H. Zhang, “Semantics-empowered communications: A tutorial-cum-survey,” IEEE Communications Surveys & Tutorials , 2023

  5. [13]

    Beyond transmitting bits: Context, semantics, and task-oriented communications,

    D. Gündüz, Z. Qin, I. E. Aguerri, H. S. Dhillon, Z. Yang, A. Yener, K. K. Wong, and C.-B. Chae, “Beyond transmitting bits: Context, semantics, and task-oriented communications,” IEEE Journal on Selected Areas in Communications, vol. 41, no. 1, pp. 5–41, 2022

  6. [14]

    Task-oriented communication design in cyber-physical systems: A survey on theory and applications,

    A. Mostaani, T. X. Vu, S. K. Sharma, V .-D. Nguyen, Q. Liao, and S. Chatzinotas, “Task-oriented communication design in cyber-physical systems: A survey on theory and applications,” IEEE Access , vol. 10, pp. 133 842–133 868, 2022

  7. [15]

    A survey on goal-oriented semantic communication: Techniques, challenges, and future directions,

    T. M. Getu, G. Kaddoum, and M. Bennis, “A survey on goal-oriented semantic communication: Techniques, challenges, and future directions,” IEEE Access, 2024

  8. [16]

    Berger, Rate Distortion Theory: A Mathematical Basis for Data Compression, ser

    T. Berger, Rate Distortion Theory: A Mathematical Basis for Data Compression, ser. Prentice-Hall electrical engineering series. Prentice- Hall, 1971

  9. [17]

    The information bottleneck method,

    N. Tishby, F. C. Pereira, and W. Bialek, “The information bottleneck method,” arXiv preprint physics/0004057 , 2000

  10. [18]

    An introduction to topological data analysis: fundamental and practical aspects for data scientists,

    F. Chazal and B. Michel, “An introduction to topological data analysis: fundamental and practical aspects for data scientists,” Frontiers in artificial intelligence, vol. 4, p. 667963, 2021

  11. [19]

    PersLay: A neural network layer for persistence diagrams and new graph topological signatures,

    M. Carrière, F. Chazal, Y . Ike, T. Lacombe, M. Royer, and Y . Umeda, “PersLay: A neural network layer for persistence diagrams and new graph topological signatures,” in International Conference on Artificial Intelligence and Statistics . PMLR, 2020, pp. 2786–2796

  12. [20]

    Topological autoen- coders,

    M. Moor, M. Horn, B. Rieck, and K. Borgwardt, “Topological autoen- coders,” in International conference on machine learning. PMLR, 2020, pp. 7045–7054

  13. [21]

    A topology layer for machine learning,

    R. B. Gabrielsson, B. J. Nelson, A. Dwaraknath, and P. Skraba, “A topology layer for machine learning,” in International Conference on Artificial Intelligence and Statistics . PMLR, 2020, pp. 1553–1563

  14. [22]

    PLLay: Efficient topological layer based on persistent landscapes,

    K. Kim, J. Kim, M. Zaheer, J. Kim, F. Chazal, and L. Wasserman, “PLLay: Efficient topological layer based on persistent landscapes,” Ad- vances in Neural Information Processing Systems , vol. 33, pp. 15 965– 15 977, 2020

  15. [23]

    Persformer: A trans- former architecture for topological machine learning,

    R. Reinauer, M. Caorsi, and N. Berkouk, “Persformer: A trans- former architecture for topological machine learning,” arXiv preprint arXiv:2112.15210, 2021

  16. [24]

    Position paper: Chal- lenges and opportunities in topological deep learning,

    T. Papamarkou, T. Birdal, M. Bronstein, G. Carlsson, J. Curry, Y . Gao, M. Hajij, R. Kwitt, P. Liò, P. Di Lorenzo et al., “Position paper: Chal- lenges and opportunities in topological deep learning,” arXiv preprint arXiv:2402.08871, 2024

  17. [25]

    Persistence-based pooling for shape pose recognition,

    T. Bonis, M. Ovsjanikov, S. Oudot, and F. Chazal, “Persistence-based pooling for shape pose recognition,” in Computational Topology in Image Context: 6th International Workshop, CTIC 2016, Marseille, France, June 15-17, 2016, Proceedings 6 . Springer, 2016, pp. 19–29

  18. [26]

    Sliced Wasserstein kernel for persistence diagrams,

    M. Carriere, M. Cuturi, and S. Oudot, “Sliced Wasserstein kernel for persistence diagrams,” in International conference on machine learning. PMLR, 2017, pp. 664–673

  19. [27]

    Time series classification via topological data analysis,

    A. Karan and A. Kaygun, “Time series classification via topological data analysis,” Expert Systems with Applications , vol. 183, p. 115326, 2021

  20. [28]

    Large scale computation of means and clusters for persistence diagrams using optimal transport,

    T. Lacombe, M. Cuturi, and S. Oudot, “Large scale computation of means and clusters for persistence diagrams using optimal transport,” Advances in Neural Information Processing Systems , vol. 31, 2018

  21. [29]

    Persistence-based segmentation of deformable shapes,

    P. Skraba, M. Ovsjanikov, F. Chazal, and L. Guibas, “Persistence-based segmentation of deformable shapes,” in 2010 IEEE Computer Society xiii Conference on Computer Vision and Pattern Recognition-Workshops . IEEE, 2010, pp. 45–52

  22. [30]

    Deep learning enabled semantic communication systems,

    H. Xie, Z. Qin, G. Y . Li, and B.-H. Juang, “Deep learning enabled semantic communication systems,” IEEE Transactions on Signal Pro- cessing, vol. 69, pp. 2663–2675, 2021

  23. [31]

    Semantic commu- nication with adaptive universal transformer,

    Q. Zhou, R. Li, Z. Zhao, C. Peng, and H. Zhang, “Semantic commu- nication with adaptive universal transformer,” IEEE Wireless Communi- cations Letters, vol. 11, no. 3, pp. 453–457, 2021

  24. [32]

    Wireless deep video semantic transmission,

    S. Wang, J. Dai, Z. Liang, K. Niu, Z. Si, C. Dong, X. Qin, and P. Zhang, “Wireless deep video semantic transmission,” IEEE Journal on Selected Areas in Communications , vol. 41, no. 1, pp. 214–229, 2022

  25. [33]

    Nonlinear transform source-channel coding for semantic communications,

    J. Dai, S. Wang, K. Tan, Z. Si, X. Qin, K. Niu, and P. Zhang, “Nonlinear transform source-channel coding for semantic communications,” IEEE Journal on Selected Areas in Communications , vol. 40, no. 8, pp. 2300– 2316, 2022

  26. [34]

    Towards image understanding from deep compression without decoding,

    R. Torfason, F. Mentzer, E. Agustsson, M. Tschannen, R. Timofte, and L. Van Gool, “Towards image understanding from deep compression without decoding,” arXiv preprint arXiv:1803.06131 , 2018

  27. [35]

    The deterministic information bottle- neck,

    D. Strouse and D. J. Schwab, “The deterministic information bottle- neck,” Neural computation, vol. 29, no. 6, pp. 1611–1630, 2017

  28. [36]

    Task-oriented image semantic communication based on rate-distortion theory,

    F. Liu, W. Tong, Y . Yang, Z. Sun, and C. Guo, “Task-oriented image semantic communication based on rate-distortion theory,” arXiv preprint arXiv:2201.10929, 2022

  29. [37]

    Learning task-oriented communication for edge inference: An information bottleneck approach,

    J. Shao, Y . Mao, and J. Zhang, “Learning task-oriented communication for edge inference: An information bottleneck approach,” IEEE Journal on Selected Areas in Communications, vol. 40, no. 1, pp. 197–211, 2021

  30. [38]

    Robust information bottleneck for task-oriented communication with digital modulation,

    S. Xie, S. Ma, M. Ding, Y . Shi, M. Tang, and Y . Wu, “Robust information bottleneck for task-oriented communication with digital modulation,” IEEE Journal on Selected Areas in Communications , vol. 41, no. 8, pp. 2577–2591, 2023

  31. [39]

    Semantic information recovery in wireless networks,

    E. Beck, C. Bockelmann, and A. Dekorsy, “Semantic information recovery in wireless networks,” Sensors, vol. 23, no. 14, p. 6347, 2023

  32. [40]

    Graph information bottle- neck,

    T. Wu, H. Ren, P. Li, and J. Leskovec, “Graph information bottle- neck,” Advances in Neural Information Processing Systems , vol. 33, pp. 20 437–20 448, 2020

  33. [41]

    Distributed variational representation learn- ing,

    I. E. Aguerri and A. Zaidi, “Distributed variational representation learn- ing,” IEEE transactions on pattern analysis and machine intelligence , vol. 43, no. 1, pp. 120–138, 2019

  34. [42]

    Perfor- mance optimization for semantic communications: An attention-based learning approach,

    Y . Wang, M. Chen, W. Saad, T. Luo, S. Cui, and H. V . Poor, “Perfor- mance optimization for semantic communications: An attention-based learning approach,” in 2021 IEEE Global Communications Conference (GLOBECOM). IEEE, 2021, pp. 1–6

  35. [43]

    Knowledge graph embed- ding: A survey of approaches and applications,

    Q. Wang, Z. Mao, B. Wang, and L. Guo, “Knowledge graph embed- ding: A survey of approaches and applications,” IEEE transactions on knowledge and data engineering , vol. 29, no. 12, pp. 2724–2743, 2017

  36. [44]

    A survey on knowledge graphs: Representation, acquisition, and applications,

    S. Ji, S. Pan, E. Cambria, P. Marttinen, and S. Y . Philip, “A survey on knowledge graphs: Representation, acquisition, and applications,” IEEE transactions on neural networks and learning systems , vol. 33, no. 2, pp. 494–514, 2021

  37. [45]

    Cognitive semantic communication systems driven by knowledge graph,

    F. Zhou, Y . Li, X. Zhang, Q. Wu, X. Lei, and R. Q. Hu, “Cognitive semantic communication systems driven by knowledge graph,” in ICC 2022-IEEE International Conference on Communications. IEEE, 2022, pp. 4860–4865

  38. [46]

    Reasoning on the air: An im- plicit semantic communication architecture,

    Y . Xiao, Y . Li, G. Shi, and H. V . Poor, “Reasoning on the air: An im- plicit semantic communication architecture,” in 2022 IEEE International Conference on Communications Workshops (ICC Workshops) . IEEE, 2022, pp. 289–294

  39. [47]

    Joint semantic-native communication and inference via minimal sim- plicial structures,

    Q. Zhao, H. Zou, M. Bennis, M. Debbah, E. Almazrouei, and F. Bader, “Joint semantic-native communication and inference via minimal sim- plicial structures,” in GLOBECOM 2023-2023 IEEE Global Communi- cations Conference. IEEE, 2023, pp. 2233–2238

  40. [48]

    Semantic- native communication: A simplicial complex perspective,

    Q. Zhao, M. Bennis, M. Debbah, and D. B. Da Costa, “Semantic- native communication: A simplicial complex perspective,” in 2022 IEEE Globecom Workshops (GC Wkshps) . IEEE, 2022, pp. 1513–1518

  41. [49]

    Neuro-symbolic artificial intelligence (ai) for intent based semantic communication,

    C. K. Thomas and W. Saad, “Neuro-symbolic artificial intelligence (ai) for intent based semantic communication,” in GLOBECOM 2022-2022 IEEE Global Communications Conference. IEEE, 2022, pp. 2698–2703

  42. [50]

    Semantic communi- cations in networked systems: A data significance perspective,

    E. Uysal, O. Kaya, A. Ephremides, J. Gross, M. Codreanu, P. Popovski, M. Assaad, G. Liva, A. Munari, B. Soret et al. , “Semantic communi- cations in networked systems: A data significance perspective,” IEEE Network, vol. 36, no. 4, pp. 233–240, 2022

  43. [51]

    G. Gan, C. Ma, and J. Wu, Data clustering: theory, algorithms, and applications. SIAM, 2020

  44. [52]

    C. M. Bishop and N. M. Nasrabadi, Pattern recognition and machine learning. Springer, 2006, vol. 4, no. 4

  45. [53]

    Logical formalizations of commonsense reasoning: a survey,

    E. Davis, “Logical formalizations of commonsense reasoning: a survey,” Journal of Artificial Intelligence Research , vol. 59, pp. 651–723, 2017

  46. [54]

    Toward causal representation learning,

    B. Schölkopf, F. Locatello, S. Bauer, N. R. Ke, N. Kalchbrenner, A. Goyal, and Y . Bengio, “Toward causal representation learning,” Proceedings of the IEEE , vol. 109, no. 5, pp. 612–634, 2021

  47. [55]

    Edelsbrunner and J

    H. Edelsbrunner and J. L. Harer, Computational topology: an introduc- tion. American Mathematical Society, 2022

  48. [56]

    Topological estimation using witness complexes

    V . De Silva and G. E. Carlsson, “Topological estimation using witness complexes.” in PBG, 2004, pp. 157–166

  49. [57]

    R. G. Gallager et al., Principles of digital communication . Cambridge University Press Cambridge, UK, 2008, vol. 1

  50. [58]

    The MNIST database of handwritten digit images for machine learning research,

    L. Deng, “The MNIST database of handwritten digit images for machine learning research,” IEEE Signal Processing Magazine, vol. 29, no. 6, pp. 141–142, 2012

  51. [59]

    Data-for-structural-semantics-using-tda,

    ICON, “Data-for-structural-semantics-using-tda,” https://github.com/ ICONgroupCWC/Data-for-Structural-Semantics-using-TDA.git, 2024

  52. [60]

    Hastie, R

    T. Hastie, R. Tibshirani, J. H. Friedman, and J. H. Friedman, The elements of statistical learning: data mining, inference, and prediction . Springer, 2009, vol. 2

  53. [61]

    T. M. Cover and J. A. Thomas, Elements of Information Theory (Wiley Series in Telecommunications and Signal Processing) . USA: Wiley- Interscience, 2006

  54. [62]

    Lin and D

    S. Lin and D. J. Costello, Error Control Coding , 2nd ed. Pearson Prentice Hall, 2004

  55. [63]

    GUDHI Editorial Board, 2020

    The GUDHI Project, GUDHI User and Reference Manual , 3.1.1 ed. GUDHI Editorial Board, 2020. [Online]. Available: https://gudhi.inria. fr/doc/3.1.1/

  56. [64]

    Pytorch: An imperative style, high-performance deep learning library,

    A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga et al. , “Pytorch: An imperative style, high-performance deep learning library,” Advances in neural information processing systems , vol. 32, 2019

  57. [65]

    Tensorflow: Large-scale machine learning on heterogeneous systems,

    M. Abadi, A. Agarwal, P. Barham, E. Brevdo, Z. Chen, C. Citro, G. S. Corrado, A. Davis, J. Dean, M. Devin et al. , “Tensorflow: Large-scale machine learning on heterogeneous systems,” 2015

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.