Pith. sign in

REVIEW 3 major objections 6 minor 95 references

Language Model for Large-Text Transmission in Noisy Quantum Communications

T0 review · 3 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read A language model plugged in after the decoder cuts noisy-quantum sentence errors from 76% to 24%.

desk verdict A real but modest error-reduction result for BERT-corrected superdense coding, undercut by an unfair classical baseline that makes entanglement look more helpful than it is. read the letter →

arxiv 2504.20842 v1 pith:W3THB3DJ submitted 2025-04-29 quant-ph

Yuqi Li , Zhouhang Shi , Haitao Ma , Li Shen , Jinge Bao , Yunlong Xiao This is my paper · ORCID
classification quant-ph MSC 81P6868T50 PACS 03.67.Hk03.67.-a
keywords superdensecodingquantumcommunicationBERTnaturallanguageprocessingerrormitigationnoisesentenceratequdit
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 proposes that the most practical way to protect large text messages sent through noisy quantum channels is not more quantum hardware but a classical language model placed after the decoder. It builds PQC-BERT, a BERT-based post-processing module, around superdense coding: text is ASCII-encoded into bits, sent through a noisy channel, decoded, then repaired first by a dictionary and then by a context-aware Transformer. In numerical simulations with bit-flip noise at strength 0.01, sentence error rates on two text corpora drop from about 76% and 73% to about 24% and 26%, with no quantum error correction or repeated sampling. The same module also works across phase-flip, depolarizing, and amplitude-damping noise without being told which noise model is present, and it performs best when messages are segmented into 5 to 16 word units. The broader claim is that quantum capacity gains and linguistic redundancy can combine to make noisy quantum text transmission practical.

What carries the argument

The load-bearing object is PQC-BERT, a two-module post-decoding pipeline. The word-level repairing module compares each decoded word against a dictionary and replaces it with the dictionary word whose ASCII bit string is closest in Hamming distance; the sentence-level repairing module then passes the repaired text through a BERT-based correction network that proposes context-appropriate words, and an evaluation network trained with focal loss decides whether each proposal should be accepted or the dictionary repair kept. The mechanism works because natural language is statistically redundant: even when individual bits are corrupted, the surrounding words strongly constrain what the damaged word must be, so a bidirectional Transformer can infer the intended text. The paper's additional claim that quantum resources improve the pipeline rests on comparing error rates after the same post-processor for classical bits, qubit superdense coding, and qudit superdense coding.

What would settle it

Send one half of a Bell state through an experimentally realized amplitude-damping channel, perform a complete two-qubit Bell measurement, and compare the outcome distribution with the one produced by the paper's Appendix C simulation of the same channel at the same noise strength; if the two distributions differ substantially, the reported gains for amplitude damping are an artifact of the simplified noise model.

Watch

Extended reading notes

Core claim

The central claim is that a classical NLP post-decoder, PQC-BERT, successfully corrects nearly all of the semantically relevant errors in text transmitted by noisy superdense coding, reducing sentence error rates from roughly 76% to below 26% at a 1% bit-flip noise level, and that it does so without the extra qubits of quantum error correction or the repeated rounds of quantum error mitigation. A further claim is that entanglement itself helps the language model: under identical bit-flip noise and noise strength, quantum superdense coding with PQC-BERT outperforms a classical bit channel with the same post-processor, and qudit superdense coding with d = 4 outperforms qubit superdense coding. The paper interprets this as showing that increased entanglement improves transmission performance even though PQC-BERT never interacts with the quantum states directly.

Load-bearing premise

The reported error reductions rest on simulating noisy channels as probabilistic bit and phase flips on randomly chosen qubits rather than as full quantum-channel evolution; if that simplified model does not reproduce the measurement statistics of real amplitude damping or depolarizing channels, the corresponding error-rate numbers would change.

Editorial extensions

If this is right

  • If the central claim holds, a noisy quantum channel can carry intelligible prose at useful rates without the physical-qubit overhead of quantum error correction or the sample overhead of quantum error mitigation.
  • Because PQC-BERT is trained without noise-model labels, a single deployment could be dropped into channels whose noise is unknown or time-varying, removing a separate noise-characterization step.
  • The reported qudit advantage implies that moving from qubits to higher-dimensional entangled systems improves not only capacity but also the effectiveness of the language-model cleanup, which would make qudit superdense coding an attractive target for implementation.
  • The 5 to 16 word optimal range suggests a concrete transmission protocol design: split long text into units of at most 16 words before sending, so that each unit is corrected in the regime where the model works best.
  • The reported error-detection accuracy suggests PQC-BERT could be used as a context-sensitive pre-screen that flags likely error locations before a conventional quantum error correction code acts on them.

Reading between the lines

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

  • A natural extension the paper leaves implicit is to test whether the same post-decoding cleanup transfers to other redundant data types, such as speech phonemes, images with spatial autocorrelation, or protocol headers, where a learned prior could substitute for the language prior used here.
  • If the entanglement advantage observed in simulation is real, the mechanism is worth isolating: it may come from the specific error patterns superdense coding produces rather than from an intrinsic quantum benefit, and a controlled experiment varying only the channel use under the same noise floor could separate those effects.
  • The numerical setup as described suggests a cheap additional test: feed the same corrupted bit strings through a strong neural decoder without any quantum protocol to see how much of the reported gain is linguistic redundancy alone and how much is specific to entanglement-assisted error patterns.
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 / 6 minor

Summary. The paper proposes PQC-BERT, a post-decoding language-model module for noisy quantum superdense coding. Text is converted to ASCII, encoded into Bell states, transmitted through noisy quantum channels (bit-flip, phase-flip, depolarizing, amplitude damping), and then repaired by a word-level dictionary module and a BERT-based sentence-level correction/evaluation network. The authors report numerical experiments on two text datasets showing that PQC-BERT substantially reduces bit, word, and sentence error rates after transmission, and they claim that entanglement-assisted (quantum) communication improves PQC-BERT's correction performance compared with a classical channel at the same noise strength.

Significance. If the results hold, the paper demonstrates a potentially useful hybrid classical-quantum approach: using a pretrained language model as a software-only postprocessor to repair text corrupted by quantum channel noise, without the qubit overhead of quantum error correction or the sampling overhead of quantum error mitigation. The core before/after error reduction is supported by held-out test data and replicated experiments, and the protocol is simple to describe. However, the headline claim that quantum entanglement improves PQC-BERT's error mitigation is undermined by an apples-to-oranges comparison, and the amplitude-damping simulation is not faithful to the stated channel. These issues are load-bearing for the paper's central message, so the results as presented cannot be accepted at face value.

major comments (3)
  1. [Sec. III.B, Table I] The classical-vs-quantum comparison is not normalized to the same per-information-bit error rate. In qubit superdense coding, a bit-flip error on the transmitted qubit acts as X⊗I on the encoded Bell state |Φ_zx⟩; since X Z^z X^x = (-1)^z Z^z X^{x⊕1}, the z bit never flips and only the x bit flips with probability λ. The effective per-bit error rate for the quantum channel is therefore λ/2, while the classical baseline has per-bit error rate λ. At λ=0.01, the quantum text entering PQC-BERT is already half as noisy as the classical text, so the higher Accuracy/F1 in Table I and the lower SER in Fig. 7 are the expected consequence of sending two bits per channel use, not evidence that entanglement improves PQC-BERT. To support the claim that 'entanglement does indeed improve PQC-BERT's performance', the comparison must be repeated with a classical baseline set to the same effective per-bit error rate (e.g., λ/2) or with an explicit resource-normalized analysis.
  2. [Appendix C and Fig. 4] The simulation of amplitude damping is described only as 'applying probabilistic noise to randomly selected qubits'. The amplitude-damping channel defined in Table II has Kraus operators K0 = diag(1, sqrt(1-λ)) and K1 = [[0, sqrt(λ)], [0,0]], which is not a Pauli channel. A random bit/phase flip does not reproduce the correct action on Bell states (for instance, K1 maps |1⟩ to |0⟩, changing the population and producing correlated errors that differ from Pauli flips). Without a faithful CPTP-map simulation or an explicit demonstration that the Bell-measurement error statistics coincide with those of the simplified model, the amplitude-damping entries in Fig. 4 and the claim of robustness across 'four distinct quantum noise models' are not supported. Please either implement the correct channel for all four models or restrict the claims to the channels actually simulated.
  3. [Sec. III.C, Fig. 7 and Appendix E] The qubit-vs-qudit comparison likewise lacks per-information-bit normalization. A qudit of dimension d = 4 transmits log2(4^2)=4 classical bits per channel use, twice as many as a qubit, and the qudit bit-flip noise defined in Table II has a different effect per encoded bit than either the qubit or the classical baseline. Without normalizing the noise strength to the probability of error per transmitted information bit (or specifying a fair cost per channel use), the conclusion that 'increased entanglement leads to improved transmission performance' is not established. The authors should re-analyze the data under a matched per-bit error rate, or clearly state that the advantage is only in raw channel-use efficiency, not in per-bit error resilience.
minor comments (6)
  1. [Sec. II.C.1] The subsection heading reads 'Word-Level Repairing Module (SLRM)', but the acronym defined and used elsewhere is WLRM; please correct the heading.
  2. [Abstract and Sec. III.C] The phrase 'successfully corrects nearly all of the errors' overstates the numerical results: after PQC-BERT, the sentence error rate remains around 24% at λ=0.01 (Sec. III.C and Fig. 4). Please align the wording with the reported SER values.
  3. [Sec. III.D] The claim that PQC-BERT operates 'without prior knowledge of the noise model or parameters' is only partially demonstrated. The training data are corrupted by the same noise models and apparently the same parameter range used in testing; please clarify what variation of noise parameters and models appears in the training set to justify the 'arbitrary noise conditions' wording.
  4. [Table I and Appendix D] The metrics Accuracy, Precision, Recall, and F1 are defined at the sentence level, while Sec. III.B discusses word-level correction. Please clarify the evaluation granularity to avoid confusion when interpreting Table I.
  5. [Fig. 8 / Appendix F] The explanation that the character 'x' in the decoded text is a generic placeholder for any non-alphabetic error character is important for interpreting the example; consider moving it to the main text near the figure instead of only in Appendix F.
  6. [References] Reference [46] (an integrated space-to-ground quantum communication network) appears in the list of sentiment-analysis BERT works; please verify that all citations in Sec. I are placed under the correct claims.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the SER-reduction claim is a supervised held-out evaluation, not a fit to its own inputs.

full rationale

The paper's central claim is that PQC-BERT, a BERT-based post-decoding module with a word-level dictionary corrector and a sentence-level correction/evaluation network, lowers sentence error rates on text corrupted by simulated quantum noise. This is an empirical machine-learning claim, not a derivation from first principles. The training pipeline is explicit: during training, WLRM-corrected text Tw is the input and the original text T is the target (Algorithm 1, Appendix C); the data is split 80:10:10 and metrics are computed on held-out test text. Thus the reported error reduction is not equivalent to a fitted parameter or to a self-definitional quantity. The protocol does not rely on any uniqueness theorem or ansatz imported from the authors' prior work; self-citations such as refs. [4], [7], [19], [31], and [79] are background or future-direction citations and are not load-bearing. Two weaknesses exist but they are correctness issues rather than circularity. First, the classical-vs-quantum comparison in Sec. III.B and Table I fixes the same bit-flip rate lambda on the classical bit channel and on the transmitted qubit; for superdense coding a single-qubit bit-flip changes the Bell outcome zx to z(x XOR 1) with probability lambda, leaving the z bit untouched, so the per-information-bit error rate is lambda/2 for the quantum path and lambda for the classical path. The higher quantum F1 and lower SER therefore reflect an unnormalized baseline, not a demonstrated entanglement-assisted enhancement of PQC-BERT; this is a benchmark-validity problem, not a circular derivation. Second, the Appendix C amplitude-damping simulation is described as probabilistic flips on randomly selected qubits rather than the CPTP map in Table II, which undermines trust in that channel's numbers but is again a modeling and credibility issue. The paper's own Discussion limits the results to simulated noise, saying the datasets are 'masked with simulated quantum noise,' which is consistent with a non-circular but simulation-only claim.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

The central claim rests on two modeling assumptions that the paper does not justify quantitatively: the noise simulation is a simplified bit-flip model rather than full state evolution, and the classical baseline is a per-bit channel that gives the quantum protocol a built-in advantage from superdense coding's two-bit capacity.

free parameters (4)
  • Focal loss alpha = not reported
    Tuning parameter in the evaluation network loss (Eq. C2), chosen to handle class imbalance.
  • Focal loss gamma = not reported
    Focusing parameter in Eq. C2.
  • Loss combination weight theta = not reported
    Weight balancing correction and evaluation losses (Eq. C3).
  • BERT fine-tuning learning rate = based on prior work [85]
    The authors state hyperparameters and learning rates were established based on prior work, so exact values are not given.
assumptions (3)
  • domain assumption Natural language text has sufficient statistical redundancy for a pre-trained language model to predict and correct corrupted words.
    The entire SLRM component assumes BERT's contextual predictions are more reliable than the noisy received text for English sentences.
  • ad hoc to paper The noise simulation, which applies probabilistic bit/phase flips to randomly selected qubits, faithfully reproduces the four quantum channels (bit-flip, phase-flip, depolarizing, amplitude damping) defined in Appendix A.
    Appendix C describes the simulation as probabilistic noise on qubits, which is accurate for Pauli channels but questionable for amplitude damping, which requires full state evolution to capture correct measurement statistics.
  • ad hoc to paper The classical communication baseline is an independent per-bit bit-flip channel with the same noise probability as the per-qubit channel.
    This baseline choice gives quantum superdense coding a built-in advantage because one qubit carries two bits, so a qubit flip at rate lambda corrupts only half as many classical bits per channel use.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Language Model for Large-Text Transmission in Noisy Quantum Communications." pith.science (2026). https://pith.science/paper/W3THB3DJ

@misc{pith2026250420842,
  author       = {Pith},
  title        = {Pith review of: Language Model for Large-Text Transmission in Noisy Quantum Communications},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/W3THB3DJ}},
  note         = {Machine review of arXiv:2504.20842}
}
read the original abstract

Quantum communication has the potential to revolutionize information processing, providing unparalleled security and increased capacity compared to its classical counterpart by using the principles of quantum mechanics. However, the presence of noise remains a major barrier to realizing these advantages. While strategies like quantum error correction and mitigation have been developed to address this challenge, they often come with substantial overhead in physical qubits or sample complexity, limiting their practicality for large-scale information transfer. Here, we present an alternative approach: applying machine learning frameworks from natural language processing to enhance the performance of noisy quantum communications, focusing on superdense coding. By employing bidirectional encoder representations from transformers (BERT), a model known for its capabilities in natural language processing, we demonstrate improvements in information transfer efficiency without resorting to conventional error correction or mitigation techniques. These results mark a step toward the practical realization of a scalable and resilient quantum internet.

Figures

Figures reproduced from arXiv: 2504.20842 by the authors.

Figure 1
Figure 1. FIG. 1 [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2 [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3 [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: FIG. 4 [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5 [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6 [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: FIG. 7 [PITH_FULL_IMAGE:figures/full_fig_p013_7.png]
Figure 8
Figure 8. Figure 8: FIG. 8 [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

95 extracted references · 57 canonical work pages

  1. [1]

    Athirdpartypreparedthe maximallyentangledstate |Φ00⟩anddistributedittothe sender, Alice, and the receiver, Bob

    EntanglementDistribution. Athirdpartypreparedthe maximallyentangledstate |Φ00⟩anddistributedittothe sender, Alice, and the receiver, Bob

  2. [2]

    Alice encodes two bits of classical informa- tion zx∈{ 0, 1}2 into one of the Bell states|Φzx⟩ by applying Pauli operationsZ(z)X(x) to her qubit

    Encoding. Alice encodes two bits of classical informa- tion zx∈{ 0, 1}2 into one of the Bell states|Φzx⟩ by applying Pauli operationsZ(z)X(x) to her qubit

  3. [3]

    Alice then sends her qubit to Bob through the noisy quantum channelE

    Noisy Communication. Alice then sends her qubit to Bob through the noisy quantum channelE

  4. [4]

    ASCII encoding

    Decoding. Upon receiving the qubit from Alice, Bob performs a Bell measurement. Based on the measure- mentoutcome,Bobdecodestheclassicalinformation zx that Alice encoded. A schematic representation of the superdense coding proto- colisprovidedinFig.1(b). BysubstitutingthePauligateswith Heisenberg-Weyloperators,wecanreadilyextendtheprotocol fromthequbitcas...

  5. [5]

    ASCII encoding

    Word-Level Repairing Module (SLRM) We begin by introducing our word-level repairing module (WLRM).Foragivenword wq i, derivedfromsuperdensecod- ing and subsequently mapped through the inverse mapping A−1 of the “ASCII encoding”, we assess its validity against thepredefineddictionary Dict. Ifwq i isfoundin Dict,werede- finewq i asww i and proceed to the ne...

  6. [6]

    Thecorrection network utilizes BERT [34] to analyze the linguistic context effectively

    Sentence-Level Repairing Module (SLRM) Oursentence-levelrepairingmodule(SLRM)isconstructed using BERT and comprises two interconnected networks: the correctionnetworkandtheevaluationnetwork. Thecorrection network utilizes BERT [34] to analyze the linguistic context effectively. However, we have observed that BERT tends to recommendchangesevenwhentheorigin...

  7. [7]

    Specifically, we consider the ideal text,T , a quote attributed to Albert Einstein: “Imagination is more FIG

    Text Transmission through Noisy Channels Here,weemployPQC-BERTtotransmitatextandevaluate its performance. Specifically, we consider the ideal text,T , a quote attributed to Albert Einstein: “Imagination is more FIG. 2. Efficient Text Transmission with PQC-BERT.In each round of noisy communication, errors manifest in differ- entways. Wepresenttwoexamples,w...

  8. [8]

    Thismetricquantifiestheratioof erroneous bits to the total number of bits transmitted

    BitErrorRate(BER). Thismetricquantifiestheratioof erroneous bits to the total number of bits transmitted. It is affected by the channel’s noise parameters, providing valuable insights into the system’s sensitivity to noise, its interference suppression capabilities, and its overall communication capacity – making it a crucial metric in information theory

Show all 95 references
  1. [9]

    Thismetricassessesthepro- portionoferrorswithintransmitteddatablocks, defined as the number of erroneous words divided by the total numberoftransmittedwords

    WordErrorRate(WER). Thismetricassessesthepro- portionoferrorswithintransmitteddatablocks, defined as the number of erroneous words divided by the total numberoftransmittedwords. Iteffectivelyevaluatesthe quality of transmission at the word level

  2. [10]

    ASCII encoding

    Sentence Error Rate (SER).This metric measures the performanceofourtextcommunicationprotocolbycal- culating the ratio of erroneous sentences to the total number of sentences transmitted. It provides valuable insightintotheprotocol’saccuracyatthesentencelevel. Figure 3 visualiz...

  3. [11]

    Bouwmeester, J

    D. Bouwmeester, J. W. Pan, K. Mattle, and et al., Experimental quantum teleportation, Nature390, 575 (1997)

    1997

  4. [12]

    Gisin and R

    N. Gisin and R. Thew, Quantum communication, Nature Pho- tonics1, 165 (2007)

    2007

  5. [13]

    Cozzolino, B

    D. Cozzolino, B. Da Lio, D. Bacco, and L. K. Oxenløwe, High- dimensional quantum communication: Benefits, progress, and future challenges, Quantum Science and Technology5, 043001 (2019)

    2019

  6. [14]

    J.Xing, Y.Li,D.Qu,L.Xiao,Z.Fan,H.Ma,P.Xue, K.Bharti, D.E.Koh,andY.Xiao,Teleportationwithembezzlingcatalysts, Communications Physics7, 357 (2024)

    2024

  7. [15]

    Hu, C.-X

    X.-M. Hu, C.-X. Huang, Y.-B. Sheng, L. Zhou, B.-H. Liu, Y. Guo, C. Zhang, W.-B. Xing, Y.-F. Huang, C.-F. Li, and G.- C. Guo, Long-distance entanglement purification for quantum communication, Phys. Rev. Lett.126, 010503 (2021)

    2021

  8. [16]

    X.-M. Hu, C. Zhang, B.-H. Liu, et al. , Experimental high- dimensionalquantumteleportation,Phys.Rev.Lett. 125,230501 (2020)

    2020

  9. [17]

    J. Xing, T. Feng, Z. Fan, H. Ma, K. Bharti, D. E. Koh, and Y.Xiao,Fundamentallimitationsoncommunicationoveraquan- tum network (2023), arXiv:2306.04983 [quant-ph]

    2023 arXiv

  10. [18]

    J.-G. Ren, P. Xu, H.-L. Yong, L. Zhang, S.-K. Liao, J. Yin, W.-Y. Liu, W.-Q. Cai, M. Yang, L. Li,et al., Ground-to-satellite quantum teleportation, Nature549, 70 (2017)

    2017

  11. [19]

    Liao, W.-Q

    S.-K. Liao, W.-Q. Cai, W.-Y. Liu, L. Zhang, Y. Li, J.-G. Ren, et al.,Satellite-to-groundquantumkeydistribution,Nature 549, 43 (2017)

    2017

  12. [20]

    C.-Y. Lu, Y. Cao, C.-Z. Peng, and J.-W. Pan, Micius quantum experiments in space, Rev. Mod. Phys.94, 035001 (2022)

    2022

  13. [21]

    Li, W.-Q

    Y. Li, W.-Q. Cai, J.-G. Ren, ,et al., Microsatellite-based real- time quantum key distribution, Nature640, 47 (2025)

    2025

  14. [22]

    Yin, J.-G

    J. Yin, J.-G. Ren, H. Lu,et al., Quantum teleportation and en- tanglementdistributionover100-kilometrefree-spacechannels, Nature488, 185 (2012)

    2012

  15. [23]

    J. L. Liu, X. Y. Luo, Y. Yu, and et al., Creation of mem- ory–memory entanglement in a metropolitan quantum network, Nature629, 579 (2024)

    2024

  16. [24]

    C. M. Knaut, A. Suleymanzade, Y. C. Wei, and et al., Entan- glement of nanophotonic quantum memory nodes in a telecom network, Nature629, 573 (2024)

    2024

  17. [25]

    C. H. Bennett, P. W. Shor, J. A. Smolin, and A. V. Thapliyal, Entanglement-assistedclassicalcapacityofnoisyquantumchan- nels, Phys. Rev. Lett.83, 3081 (1999)

    1999

  18. [26]

    Harrow, P

    A. Harrow, P. Hayden, and D. Leung, Superdense coding of quantum states, Phys. Rev. Lett.92, 187901 (2004)

    2004

  19. [27]

    J. T. Barreiro, T.-C. Wei, and P. G. Kwiat, Beating the chan- nel capacity limit for linear photonic superdense coding, Nature Physics4, 282 (2008)

    2008

  20. [28]

    X.-M. Hu, Y. Guo, B.-H. Liu, Y.-F. Huang, C.-F. Li, and G.-C. Guo, Beating the channel capacity limit for superdense coding with entangled ququarts, Science Advances4, eaat9304 (2018)

    2018

  21. [29]

    Y. Li, J. Xing, D. Qu, L. Xiao, Z. Fan, Z.-J. Zheng, H. Ma, P.Xue,K.Bharti,D.E.Koh,andY.Xiao,Communicationwith quantum catalysts (2024), arXiv:2406.14395 [quant-ph]

    2024

  22. [30]

    Knill and R

    E. Knill and R. Laflamme, Theory of quantum error-correcting codes, Phys. Rev. A55, 900 (1997)

    1997

  23. [31]

    T. Aoki, G. Takahashi, T. Kajiya, and et al., Quantum error correction beyond qubits, Nature Physics5, 541 (2009)

    2009

  24. [32]

    B. M. Terhal, Quantum error correction for quantum memories, Rev. Mod. Phys.87, 307 (2015)

    2015

  25. [33]

    Krinner, N

    S. Krinner, N. Lacroix, A. Remm, and et al., Realizing repeated quantumerrorcorrectioninadistance-threesurfacecode,Nature 605, 669 (2022)

    2022

  26. [34]

    V. V. Sivak, A. Eickbusch, B. Royer, and et al., Real-time quan- tum error correction beyond break-even, Nature616, 50 (2023)

    2023

  27. [35]

    Liu and S

    Z.-W. Liu and S. Zhou, Approximate symmetries and quantum error correction, npj Quantum Information9, 119 (2023)

    2023

  28. [36]

    Z.-W.LiuandS.Zhou,Ageneraltheoryofquantumcodescon- necting quantum computation, complexity and physics, Nature Physics20, 1708 (2024)

    2024

  29. [37]

    S. Endo, S. C. Benjamin, and Y. Li, Practical quantum error mitigation for near-future applications, Phys. Rev. X8, 031027 (2018)

    2018

  30. [38]

    Z. Cai, R. Babbush, S. C. Benjamin, S. Endo, W. J. Huggins, Y. Li, J. R. McClean, and T. E. O’Brien, Quantum error mitiga- tion, Rev. Mod. Phys.95, 045005 (2023)

    2023

  31. [39]

    Takagi, H

    R. Takagi, H. Tajima, and M. Gu, Universal sampling lower bounds for quantum error mitigation, Phys. Rev. Lett. 131, 210602 (2023)

    2023

  32. [40]

    Takagi, S

    R. Takagi, S. Endo, S. Minagawa, and M. Gu, Fundamental limits of quantum error mitigation, npj Quantum Information8, 11 (2022)

    2022

  33. [41]

    Z. Liu, Y. Xiao, and Z. Cai, Non-markovian noise sup- pression simplified through channel representation (2024), arXiv:2412.11220 [quant-ph]

    2024 arXiv

  34. [42]

    M. Liao, Y. Zhu, G. Chiribella, and Y. Yang, Noise-agnostic quantum error mitigation with data augmented neural models, npj Quantum Information11, 8 (2025)

    2025

  35. [43]

    Gomez,L.u.Kaiser,andI.Polosukhin,Attentionisallyouneed, Advances in Neural Information Processing Systems30(2017)

    A.Vaswani,N.Shazeer,N.Parmar,J.Uszkoreit,L.Jones,A.N. Gomez,L.u.Kaiser,andI.Polosukhin,Attentionisallyouneed, Advances in Neural Information Processing Systems30(2017)

    2017

  36. [44]

    Devlin, M.-W

    J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, BERT: Pre-training of deep bidirectional transformers for language un- derstanding, Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Lin- guistics: Human Language Technologie...

    2019

  37. [45]

    Bello, S.-C

    A. Bello, S.-C. Ng, and M.-F. Leung, A bert framework to sen- timent analysis of tweets, Sensors23, 506 (2023)

    2023

  38. [46]

    Cao and C

    J. Cao and C. Lai, A bilingual multi-type spam detection model based on m-bert, GLOBECOM 2020-2020 IEEE Global Com- munications Conference , 1 (2020)

    2020

  39. [47]

    C.Oswald,S.E.Simon,andA.Bhattacharya,Spotspam: Inten- tionanalysis–drivensmsspamdetectionusingbertembeddings, ACM Transactions on the Web (TWEB)16, 1 (2022)

    2022

  40. [48]

    Hakala and S

    K. Hakala and S. Pyysalo, Biomedical named entity recogni- tion with multilingual bert, Proceedings of the 5th workshop on BioNLP open shared tasks , 56 (2019). 10

    2019

  41. [49]

    Chang, L

    Y. Chang, L. Kong, K. Jia, and Q. Meng, Chinese named entity recognitionmethodbasedonbert,2021IEEEinternationalcon- ference on data science and computer application (ICDSCA) , 294 (2021)

    2021

  42. [50]

    Z. Yang, Z. Dai, Y. Yang, J. Carbonell, R. Salakhutdinov, and Q. V. Le, Xlnet: generalized autoregressive pretraining for lan- guageunderstanding,Proceedingsofthe33rdInternationalCon- ference on Neural Information Processing Systems (2019)

    2019

  43. [51]

    Z.Liu,W.Lin,Y.Shi,andJ.Zhao,Arobustlyoptimizedbertpre- training approach with post-training, Chinese Computational LinguisticsLNCS 13175, 471 (2021)

    2021

  44. [52]

    Z. Lan, M. Chen, S. Goodman, K. Gimpel, P. Sharma, and R. Soricut, Albert: A lite bert for self-supervised learning of language representations (2020), arXiv:1909.11942 [cs.CL]

    2020 arXiv

  45. [53]

    M.Hoang,O.A.Bihorac,andJ.Rouces,Aspect-basedsentiment analysis using bert, Proceedings of the 22nd nordic conference on computational linguistics , 187 (2019)

    2019

  46. [54]

    H. Xu, B. Liu, L. Shu, and S. Y. Philip, Bert post-training for reviewreadingcomprehensionandaspect-basedsentimentanal- ysis,Proceedingsofthe2019ConferenceoftheNorthAmerican Chapter of the Association for Computational Linguistics: Hu- manLanguageTechnologies,Volume1(LongandSh...

    2019

  47. [55]

    M. G. Sousa, K. Sakiyama, L. d. S. Rodrigues, P. H. Moraes, E. R. Fernandes, and E. T. Matsubara, Bert for stock market sentimentanalysis,2019IEEE31stInternationalConferenceon Tools with Artificial Intelligence (ICTAI) , 1597 (2019)

    2019

  48. [56]

    Y. Cao, Q. Zhang, T.-Y. Chen,et al., An integrated space-to- groundquantumcommunicationnetworkover4,600kilometres, Nature589, 214 (2021)

    2021

  49. [57]

    N. J. Prottasha, A. A. Sami, M. Kowsher, S. A. Murad, A. K. Bairagi,M.Masud,andM.Baz,Transferlearningforsentiment analysis using bert based supervised fine-tuning, Sensors22, 4157 (2022)

    2022

  50. [58]

    Zhang, H

    S. Zhang, H. Huang, J. Liu, and H. Li, Spelling error correction with soft-masked bert, Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics , 882 (2020)

    2020

  51. [59]

    M. Tan, D. Chen, Z. Li, and P. Wang, Spelling error correction with bert based on character-phonetic, 2020 IEEE 6th Interna- tional Conference on Computer and Communications (ICCC) , 1146 (2020)

    2020

  52. [60]

    T. T. H. Nguyen, A. Jatowt, N.-V. Nguyen, M. Coustaty, and A. Doucet, Neural machine translation with bert for post-ocr error detection and correction, Proceedings of the ACM/IEEE joint conference on digital libraries in 2020 , 333 (2020)

    2020

  53. [61]

    Horodecki, P

    R. Horodecki, P. Horodecki, M. Horodecki, and K. Horodecki, Quantum entanglement, Rev. Mod. Phys.81, 865 (2009)

    2009

  54. [62]

    Khatri and M

    S. Khatri and M. M. Wilde, Principles of quantum communi- cation theory: A modern approach (2024), arXiv:2011.04672 [quant-ph]

    2024 arXiv

  55. [63]

    96, 190501 (2006)

    C.Schuck,G.Huber,C.Kurtsiefer,andH.Weinfurter,Complete deterministiclinearopticsbellstateanalysis,Phys.Rev.Lett. 96, 190501 (2006)

    2006

  56. [64]

    Hodosh, P

    M. Hodosh, P. Young, and J. Hockenmaier, Framing image de- scriptionasarankingtask: Data,modelsandevaluationmetrics, Journal of Artificial Intelligence Research47, 853 (2013)

    2013

  57. [65]

    Warstadt, A

    A. Warstadt, A. Singh, and S. R. Bowman, A corpus for mod- eling rich morphological agreement, Proceedings of the 2019 Conference of the North American Chapter of the Association forComputationalLinguistics: HumanLanguageTechnologies, Volume 1 (Long and Short Papers) , 4186 (2019)

    2019

  58. [66]

    Beltagy, M

    I. Beltagy, M. E. Peters, and A. Cohan, Longformer: The long- document transformer (2020), arXiv:2004.05150 [cs.CL]

    2020 arXiv

  59. [67]

    Zaheer, G

    M. Zaheer, G. Guruganesh, A. Dubey, J. Ainslie, C. Alberti, S. Ontanon, P. Pham, A. Ravula, Q. Wang, L. Yang, and A.Ahmed,Bigbird: Transformersforlongersequences(2021), arXiv:2007.14062 [cs.CL]

    2021 arXiv

  60. [68]

    Chatterjee, K

    A. Chatterjee, K. Phalak, and S. Ghosh, Quantum error cor- rection for dummies, 2023 IEEE International Conference on Quantum Computing and Engineering (QCE)01, 70 (2023)

    2023

  61. [69]

    J.Carrasquilla,Machinelearningforquantummatter,Advances in Physics: X5, 1797528 (2020)

    2020

  62. [70]

    R.G.Melko,G.Carleo,J.Carrasquilla,andJ.I.Cirac,Restricted boltzmannmachinesinquantumphysics,NaturePhysics 15,887 (2019)

    2019

  63. [71]

    G.CarleoandM.Troyer,Solvingthequantummany-bodyprob- lem with artificial neural networks, Science355, 602 (2017)

    2017

  64. [72]

    Carrasquilla and R

    J. Carrasquilla and R. G. Melko, Machine learning phases of matter, Nature Physics13, 431 (2017)

    2017

  65. [73]

    Wallnöfer, A

    J. Wallnöfer, A. A. Melnikov, W. Dür, and H. J. Briegel, Ma- chine learning for long-distance quantum communication, PRX Quantum1, 010301 (2020)

    2020

  66. [74]

    Lloyd, Quantum machine learning, Nature549, 195 (2017)

    J.Biamonte,P.Wittek,N.Pancotti,P.Rebentrost,N.Wiebe,and S. Lloyd, Quantum machine learning, Nature549, 195 (2017)

    2017

  67. [75]

    Dunjko and H

    V. Dunjko and H. J. Briegel, Machine learning & artificial in- telligence in the quantum domain: a review of recent progress, Reports on Progress in Physics81, 074001 (2018)

    2018

  68. [76]

    N. Guo, Z. Yu, M. Choi, A. Agrawal, K. Nakaji, A. Aspuru- Guzik,andP.Rebentrost,Quantumlinearalgebraisallyouneed for transformer architectures (2024), arXiv:2402.16714 [quant- ph]

    2024

  69. [77]

    Z. Yu, Q. Chen, Y. Jiao, Y. Li, X. Lu, X. Wang, and J. Z. Yang, Non-asymptotic approximation error bounds of parameterized quantum circuits (2024), arXiv:2310.07528 [quant-ph]

    2024 arXiv

  70. [78]

    E. A. Cherrat, I. Kerenidis, N. Mathur, J. Landman, M. Strahm, and Y. Y. Li, Quantum vision transformers, Quantum8, 1265 (2024)

    2024

  71. [79]

    J. Tian, X. Sun, Y. Du, S. Zhao, Q. Liu, K. Zhang, W. Yi, W. Huang, C. Wang, X. Wu, M.-H. Hsieh, T. Liu, W. Yang, and D. Tao, Recent advances for quantum neural networks in generative learning, IEEE Transactions on Pattern Analysis and Machine Intelligence45, 12321 (2023)

    2023

  72. [80]

    X. Wang, J. Liu, T. Liu, Y. Luo, Y. Du, and D. Tao, Symmetric pruning in quantum neural networks (2023), arXiv:2208.14057 [quant-ph]

    2023 arXiv

  73. [81]

    Slussarenko and G

    S. Slussarenko and G. J. Pryde, Photonic quantum informa- tion processing: A concise review, Applied Physics Reviews 6, 041303 (2019)

    2019

  74. [82]

    C.FabreandN.Treps,Modesandstatesinquantumoptics,Rev. Mod. Phys.92, 035005 (2020)

    2020

  75. [83]

    Y. Yuan, Y. Xiao, Z. Hou, S.-M. Fei, G. Gour, G.-Y. Xiang, C.- F. Li, and G.-C. Guo, Strong majorization uncertainty relations and experimental verifications, npj Quantum Information9, 65 (2023)

    2023

  76. [84]

    Rivas and S

    A. Rivas and S. F. Huelga,Open Quantum Systems: An Intro- duction (Springer, 2012)

    2012

  77. [85]

    F. A. Pollock, C. Rodríguez-Rosario, T. Frauenheim, M. Pater- nostro,andK.Modi,Non-markovianquantumprocesses: Com- plete framework and efficient characterization, Phys. Rev. A97, 012127 (2018)

    2018

  78. [86]

    D.A.Lidar,Lecturenotesonthetheoryofopenquantumsystems (2020), arXiv:1902.00967 [quant-ph]

    2020 arXiv

  79. [87]

    S.MilzandK.Modi,Quantumstochasticprocessesandquantum non-markovian phenomena, PRX Quantum2, 030201 (2021)

    2021

  80. [88]

    Y. Xiao, K. Sengupta, S. Yang, and G. Gour, Uncertainty prin- ciple of quantum processes, Phys. Rev. Res.3, 023077 (2021)

    2021

  81. [89]

    Y. Xiao, Y. Yang, X. Wang, Q. Liu, and M. Gu, Quantum un- certainty principles for measurements with interventions, Phys. Rev. Lett.130, 240201 (2023)

    2023

  82. [90]

    Taranto, S

    P. Taranto, S. Milz, M. Murao, M. T. Quintino, and K. Modi, Higher-order quantum operations (2025), arXiv:2503.09693 [quant-ph]. 11

    2025 arXiv

  83. [91]

    Gottesman, Theory of fault-tolerant quantum computation, Phys

    D. Gottesman, Theory of fault-tolerant quantum computation, Phys. Rev. A57, 127 (1998)

    1998

  84. [92]

    D. A. Lidar and T. A. Brun, eds.,Quantum Error Correction (Cambridge University Press, Cambridge, 2013)

    2013

  85. [93]

    Gottesman, Surviving as a quantum computer in a classical world, Textbook manuscript preprint (2024)

    D. Gottesman, Surviving as a quantum computer in a classical world, Textbook manuscript preprint (2024)

    2024

  86. [94]

    Ross and G

    T.-Y. Ross and G. Dollár, Focal loss for dense object detection, proceedings of the IEEE conference on computer vision and pattern recognition , 2980 (2017)

    2017

  87. [95]

    x” in the decoded text serve as generic error indicators: they do not represent corrupted instances of the letter “x

    A. J. Quijano, S. Nguyen, and J. Ordonez, Grid search hyperpa- rameter benchmarking of bert, albert, and longformer on duorc (2021), arXiv:2101.06326 [cs.CL]. Appendix A: Noise Models In a closed quantum system, the evolution of the quantum state is reversible and governed by u...

    2021 arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.