arXiv subjects

Yao Fu

Publications and source records attributed to Yao Fu.

104 records · Page 6Linked to original sources

arXiv2020

Latent Template Induction with Gumbel-CRFs

Learning to control the structure of sentences is a challenging problem in text generation. Existing work either relies on simple deterministic approaches or RL-based hard structures. We explore the use of structured variational autoencoders to infer latent templates for sentence generation using a soft, continuous relaxation in order to utilize reparameterization for training. Specifically, we propose a Gumbel-CRF, a continuous relaxation of the CRF sampling algorithm using a relaxed Forward-Filtering Backward-Sampling (FFBS) approach. As a reparameterized gradient estimator, the Gumbel-CRF gives more stable gradients than score-function based estimators. As a structured inference network, we show that it learns interpretable templates during training, which allows us to control the decoder during testing. We demonstrate the effectiveness of our methods with experiments on data-to-text generation and unsupervised paraphrase generation.

cs.CL↗

arXiv2020

Reduction of PDF uncertainty in the measurement of the weak mixing angle at the ATLAS experiment

We investigate the parton distribution function (PDF) uncertainty in the measurement of the effective weak mixing angle $\sin^2θ_{\text{eff}}^{\ell}$ at the CERN Large Hadron Collider (LHC). The PDF-induced uncertainty is large in the proton-proton collisions at the LHC due to the dilution effect. The measurement of the Drell-Yan forward-backward asymmetry ($A_{FB}$) at the LHC can be used to reduce the PDF uncertainty in the $\sin^2θ_{\text{eff}}^{\ell}$ measurement. However, when including the full mass range of lepton pairs in the $A_{FB}$ data analysis, the correlation between the PDF updating procedure and the $\sin^2θ_{\text{eff}}^{\ell}$ extraction leads to a sizable bias in the obtained $\sin^2θ_{\text{eff}}^{\ell}$ value. From our studies, we find that the bias can be significantly reduced by removing Drell-Yan events with invariant mass around the $Z$ pole region, while most of the sensitivity in reducing the PDF uncertainty remains. Furthermore, the lepton charge asymmetry in the $W$ boson events as a function of the rapidity of the charged leptons, $A_\pm(η_\ell)$, is known to be another observable which can be used to reduce the PDF uncertainty in the $\sin^2θ_{\text{eff}}^{\ell}$ measurement. The constraint from $A_\pm(η_\ell)$ is complementary to that from the $A_{FB}$, thus no bias affects the $\sin^2θ_{\text{eff}}^{\ell}$ extraction. The studies are performed using the Error PDF Updating Method Package ({\sc ePump}), which is based on the Hessian updating methods. In this article, the CT14HERA2 PDF set is used as an example.

hep-ex↗

arXiv2020

Paraphrase Generation with Latent Bag of Words

Paraphrase generation is a longstanding important problem in natural language processing. In addition, recent progress in deep generative models has shown promising results on discrete latent variables for text generation. Inspired by variational autoencoders with discrete latent structures, in this work, we propose a latent bag of words (BOW) model for paraphrase generation. We ground the semantics of a discrete latent variable by the BOW from the target sentences. We use this latent variable to build a fully differentiable content planning and surface realization model. Specifically, we use source words to predict their neighbors and model the target BOW with a mixture of softmax. We use Gumbel top-k reparameterization to perform differentiable subset sampling from the predicted BOW distribution. We retrieve the sampled word embeddings and use them to augment the decoder and guide its generation search space. Our latent BOW model not only enhances the decoder, but also exhibits clear interpretability. We show the model interpretability with regard to \emph{(i)} unsupervised learning of word neighbors \emph{(ii)} the step-by-step generation procedure. Extensive experiments demonstrate the transparent and effective generation process of this model.\footnote{Our code can be found at \url{https://github.com/FranxYao/dgm_latent_bow}}

cs.CL↗

arXiv2019

Rethinking Text Attribute Transfer: A Lexical Analysis

Text attribute transfer is modifying certain linguistic attributes (e.g. sentiment, style, authorship, etc.) of a sentence and transforming them from one type to another. In this paper, we aim to analyze and interpret what is changed during the transfer process. We start from the observation that in many existing models and datasets, certain words within a sentence play important roles in determining the sentence attribute class. These words are referred to as \textit{the Pivot Words}. Based on these pivot words, we propose a lexical analysis framework, \textit{the Pivot Analysis}, to quantitatively analyze the effects of these words in text attribute classification and transfer. We apply this framework to existing datasets and models and show that: (1) the pivot words are strong features for the classification of sentence attributes; (2) to change the attribute of a sentence, many datasets only requires to change certain pivot words; (3) consequently, many transfer models only perform the lexical-level modification, while leaving higher-level sentence structures unchanged. Our work provides an in-depth understanding of linguistic attribute transfer and further identifies the future requirements and challenges of this task\footnote{Our code can be found at https://github.com/FranxYao/pivot_analysis}.

cs.CL↗

arXiv2019

Measurement-Device-Independent Twin-Field Quantum Key Distribution

The ultimate aim of quantum key distribution (QKD) is improving the performance of transmission distance and key generation speed. Unfortunately, it is believed to be limited by the secret-key capacity of quantum channel without quantum repeater. Recently, a novel twin-field QKD (TFQKD) [Nature 557, 400 (2018)] is proposed to break through the limit, where the key rate is proportional to the square-root of channel transmittance. Here, by using the vacuum and one-photon state as a qubit, we show that the TF-QKD can be regarded as a measurement-device-independent QKD (MDI-QKD) with single-photon Bell state measurement. Therefore, the MDI property of TF-QKD can be understood clearly. Importantly, the universal security proof theories can be directly used for the TF-QKD, such as BB84, six-state and reference-frame-independent schemes. Furthermore, we propose a feasible experimental scheme for the proof-of-principle experimental demonstration.

quant-ph↗

arXiv2019

Phase self-aligned continuous-variable measurement-device-independent quantum key distribution

Continuous-variable measurement-independent-device quantum key distribution (CV-MDI-QKD) can offer high secure key rate at metropolitan distance and remove all side channel loopholes of detection as well. However, there is no complete experimental demonstration of CV-MDI-QKD due to the remote distance phase-locking techniques challenge. Here, we present a new optical scheme to overcome this difficulty and also removes the requirement of two identical independent lasers. We anticipate that our new scheme can be used to demonstrate the in-field CV-MDI-QKD experiment and build the CV-MDI-QKD network with untrusted source.

quant-ph↗

arXiv2016

Experimental Quantum Digital Signature over 102 km

Quantum digital signature (QDS) is an approach to guarantee the nonrepudiation, unforgeability and transferability of a signature with the information-theoretical security. All previous experimental realizations of QDS relied on an unrealistic assumption of secure channels and the longest distance is only several kilometers. Here, we have experimentally demonstrated a recently proposed QDS protocol without any secure channel. Exploiting the decoy state modulation, we have successfully signed one bit message through up to 102 km optical fiber. Furthermore, we continuously run the system to sign the longer message "USTC" with 32 bit at the distance of 51 km. Our results pave the way towards the practical application of QDS.

quant-ph↗

arXiv2016

Security of quantum key distribution with multiphoton components

Most qubit-based quantum key distribution (QKD) protocols extract the secure key merely from single-photon component of the attenuated lasers. However, with the Scarani-Acin-Ribordy-Gisin 2004 (SARG04) QKD protocol, the unconditionally secure key can be extracted from the two-photon component by modifying the classical post-processing procedure in the BB84 protocol. Employing the merits of SARG04 QKD protocol and six-state preparation, one can extract secure key from the components of single photon up to four photons. In this paper, we provide the exact relations between the secure key rate and the bit error rate in a six-state SARG04 protocol with single-photon, two-photon, three-photon, and four-photon sources. By restricting the mutual information between the phase error and bit error, we obtain a higher secure bit error rate threshold of the multiphoton components than previous works. Besides, we compare the performances of the six-state SARG04 with other prepare-and-measure QKD protocols using decoy states.

quant-ph↗

arXiv2016

Practical Quantum Digital Signature

Guaranteeing nonrepudiation, unforgeability as well as transferability of a signature is one of the most vital safeguards in today's e-commerce era. Based on fundamental laws of quantum physics, quantum digital signature (QDS) aims to provide information-theoretic security for this cryptographic task. However, up to date, the previously proposed QDS protocols are impractical due to various challenging problems and most importantly, the requirement of authenticated (secure) quantum channels between participants. Here, we present the first quantum digital signature protocol that removes the assumption of authenticated quantum channels while remaining secure against the collective attacks. Besides, our QDS protocol can be practically implemented over more than 100 km under current mature technology as used in quantum key distribution.

quant-ph↗

arXiv2016

Detector-decoy quantum key distribution without monitoring signal disturbance

The round-robin differential phase-shift quantum key distribution protocol provides a secure way to exchange private information without monitoring conventional disturbances and still maintains a high tolerance of noise, making it desirable for practical implementations of quantum key distribution. However, photon number resolving detectors are required to ensure that the detected signals are single photons in the original protocol. Here, we adopt the detector-decoy method and give the bounds to the fraction of detected events from single photons. Utilizing the advantages of the protocol, we provide a practical method of performing the protocol with desirable performances requiring only threshold single-photon detectors.

quant-ph↗

arXiv2015

Long-Distance Measurement-Device-Independent Multiparty Quantum Communication

The Greenberger-Horne-Zeilinger (GHZ) entanglement, originally introduced to uncover the extreme violation of local realism against quantum mechanics, is an important resource for multiparty quantum communication tasks. But the low intensity and fragility of the GHZ entanglement source in current conditions have made the practical applications of these multiparty tasks an experimental challenge. Here we propose a feasible scheme for practically distributing the post-selected GHZ entanglement over a distance of more than 100 km for experimentally accessible parameter regimes. Combining the decoy-state and measurement-device-independent protocols for quantum key distribution, we anticipate that our proposal suggests an important avenue for practical multiparty quantum communication.

quant-ph↗

arXiv2014

Long distance measurement-device-independent quantum key distribution with coherent-state superpositions

Measurement-device-independent quantum key distribution (MDI-QKD) with decoy-state method is believed to be securely applied to defeat various hacking attacks in practical quantum key distribution systems. Recently, the coherent-state superpositions (CSS) have emerged as an alternative to single-photon qubits for quantum information processing and metrology. Here, in this Letter, CSS are exploited as the source in MDI-QKD. We present an analytical method which gives two tight formulas to estimate the lower bound of yield and the upper bound of bit error rate. We exploit the standard statistical analysis and Chernoff bound to perform the parameter estimation. Chernoff bound can provide good bounds in the long distance MDI-QKD. Our results show that with CSS, both the security transmission distance and secure key rate are significantly improved compared with those of the weak coherent states in the finite-data case.

quant-ph↗

arXiv2014

Violations of entropic Bell inequalities with coarse-grained quadrature measurements for continuous-variable states

It is a long-standing belief, as pointed out by Bell in 1986, that it is impossible to use a two-mode Gaussian state possessing a positive-definite Wigner function to demonstrate nonlocality as the Wigner function itself provides a local hidden-variable model. In particular, when one performs continuous-variable (CV) quadrature measurements upon a routinely generated CV entanglement, namely, the two-mode squeezed vacuum (TMSV) state, the resulting Wigner function is positive-definite and as such, the TMSV state cannot violate any Bell inequality using CV quadrature measurements. We show here, however, that a Bell inequality for CV states in terms of entropies can be quantum mechanically violated by the TMSV state with two coarse-grained quadrature measurements per site within experimentally accessible parameter regime. The proposed CV entropic Bell inequality is advantageous for an experimental test, especially for a possible loophole-free test of nonlocality, as the quadrature measurements can be implemented with homodyne detections of nearly 100\% detection efficiency under current technology.

quant-ph↗

arXiv2014

Measurement-device-independent quantum key distribution based on Bell's inequality

We propose two quantum key distribution (QKD) protocols based on Bell's inequality, which can be considered as modified time-reversed E91 protocol. Similar to the measurement-device-independent quantum key distribution (MDI-QKD) protocol, the first scheme requires the assumption that Alice and Bob perfectly characterize the encoded quantum states. However, our second protocol does not require this assumption, which can defeat more known and unknown source-side attacks compared with the MDI-QKD. The two protocols are naturally immune to all hacking attacks with respect to detections. Therefore, the security of the two protocols can be proven based on the violation of Bell's inequality with measurement data under fair-sampling assumption. In our simulation, the results of both protocols show that long-distance quantum key distribution over 200 km remains secure with conventional lasers in the asymptotic-data case. We present a new technique to estimate the Bell's inequality violation, which can also be applied to other fields of quantum information processing.

quant-ph↗