arXiv subjects

Wei Song

Publications and source records attributed to Wei Song.

At least 163 records · Page 9Linked to original sources

arXiv2020

Improving Prosody Modelling with Cross-Utterance BERT Embeddings for End-to-end Speech Synthesis

Despite prosody is related to the linguistic information up to the discourse structure, most text-to-speech (TTS) systems only take into account that within each sentence, which makes it challenging when converting a paragraph of texts into natural and expressive speech. In this paper, we propose to use the text embeddings of the neighboring sentences to improve the prosody generation for each utterance of a paragraph in an end-to-end fashion without using any explicit prosody features. More specifically, cross-utterance (CU) context vectors, which are produced by an additional CU encoder based on the sentence embeddings extracted by a pre-trained BERT model, are used to augment the input of the Tacotron2 decoder. Two types of BERT embeddings are investigated, which leads to the use of different CU encoder structures. Experimental results on a Mandarin audiobook dataset and the LJ-Speech English audiobook dataset demonstrate the use of CU information can improve the naturalness and expressiveness of the synthesized speech. Subjective listening testing shows most of the participants prefer the voice generated using the CU encoder over that generated using standard Tacotron2. It is also found that the prosody can be controlled indirectly by changing the neighbouring sentences.

eess.AS↗

arXiv2020

Modular Hamiltonians in flat holography and (W)AdS/WCFT

We study several aspects of holographic entanglement in two models known as flat$_3$/BMSFT and (W)AdS$_3$/WCFT. These are two examples of holography beyond AdS/CFT where the the boundary field theories are not Lorentz invariant but still feature an infinite set of local symmetries. In the first example, BMS-invariant field theories (BMSFTs) are conjectured to provide a holographic description of quantum gravity in asymptotically flat three-dimensional spacetimes; while in the second example, warped conformal field theories (WCFTs) are proposed to describe quantum gravity in warped AdS$_3$ or AdS$_3$ backgrounds with Dirichlet-Neumann boundary conditions. In particular, we derive the modular Hamiltonian for single intervals in both BMSFTs and WCFTs and find the holographic duals in the bulk using the covariant formulation of gravitational charges. We also extend the first law of entanglement entropy to these models of non-AdS holography and discuss the bound on "modular chaos" introduced recently in the context of the AdS/CFT correspondence.

hep-th↗

arXiv2020

Randomized Last-Level Caches Are Still Vulnerable to Cache Side-Channel Attacks! But We Can Fix It

Cache randomization has recently been revived as a promising defense against conflict-based cache side-channel attacks. As two of the latest implementations, CEASER-S and ScatterCache both claim to thwart conflict-based cache side-channel attacks using randomized skewed caches. Unfortunately, our experiments show that an attacker can easily find a usable eviction set within the chosen remap period of CEASER-S and increasing the number of partitions without dynamic remapping, such as ScatterCache, cannot eliminate the threat. By quantitatively analyzing the access patterns left by various attacks in the LLC, we have newly discovered several problems with the hypotheses and implementations of randomized caches, which are also overlooked by the research on conflict-based cache side-channel attack. However, cache randomization is not a false hope and it is an effective defense that should be widely adopted in future processors. The newly discovered problems are corresponding to flaws associated with the existing implementation of cache randomization and are fixable. Several new defense techniques are proposed in this paper. our experiments show that all the newly discovered vulnerabilities of existing randomized caches are fixed within the current performance budget. We also argue that randomized set-associative caches can be sufficiently strengthened and possess a better chance to be actually adopted in commercial processors than their skewed counterparts as they introduce less overhaul to the existing cache structure.

cs.CR↗

arXiv2020

Entanglement Entropy in Flat Holography

BMS symmetry, which is the asymptotic symmetry at null infinity of flat spacetime, is an important input for flat holography. In this paper, we give a holographic calculation of entanglement entropy and Rényi entropy in three dimensional Einstein gravity and Topologically Massive Gravity. The geometric picture for the entanglement entropy is the length of a spacelike geodesic which is connected to the interval at null infinity by two null geodesics. The spacelike geodesic is the fixed points of replica symmetry, and the null geodesics are along the modular flow. Our strategy is to first reformulate the Rindler method for calculating entanglement entropy in a general setup, and apply it for BMS invariant field theories, and finally extend the calculation to the bulk.

hep-th↗

arXiv2020

Heating up holography for single-trace $J\bar{T}$ deformations

We study thermodynamic aspects of a tractable toy model of holography for extremal Kerr black holes proposed in [arXiv:1806.10127]. On the gravity side, the theory can be described by the worldsheet action of string theory on a warped AdS$_3$ background supported by NS-NS flux. Once we turn on temperature, the deformed background is described by a black string solution of type IIB supergravity that features a locally warped AdS$_3$ factor. The dual field theory is conjectured to be a single-trace version of a $J\bar{T}$-deformed CFT at finite temperature. As evidence for the correspondence we show that the spectrum of strings winding on the deformed background agrees with the spectrum of $J\bar{T}$-deformed CFTs. Furthermore, we show that the gravitational charges of the black string match the averaged charges of a thermal ensemble in the dual field theory. Finally, we reproduce the Bekenstein-Hawking entropy of the black string from the microscopic density of states of $J\bar{T}$-deformed CFTs.

hep-th↗

arXiv2019

Neural Simile Recognition with Cyclic Multitask Learning and Local Attention

Simile recognition is to detect simile sentences and to extract simile components, i.e., tenors and vehicles. It involves two subtasks: {\it simile sentence classification} and {\it simile component extraction}. Recent work has shown that standard multitask learning is effective for Chinese simile recognition, but it is still uncertain whether the mutual effects between the subtasks have been well captured by simple parameter sharing. We propose a novel cyclic multitask learning framework for neural simile recognition, which stacks the subtasks and makes them into a loop by connecting the last to the first. It iteratively performs each subtask, taking the outputs of the previous subtask as additional inputs to the current one, so that the interdependence between the subtasks can be better explored. Extensive experiments show that our framework significantly outperforms the current state-of-the-art model and our carefully designed baselines, and the gains are still remarkable using BERT.

cs.CL↗

arXiv2019

Structure Constants from Modularity in Warped CFT

We derive a universal formula for the asymptotic growth of the mean value of three-point coefficient for Warped Conformal Field Theories (WCFTs), and provide a holographic calculation in BTZ black holes. WCFTs are two dimensional quantum field theories featuring a chiral Virasoro and U(1) Kac-Moody algebra, and are conjectured to be holographically dual to quantum gravity on asymptotically AdS$_3$ spacetime with Comp$\grave{\mathrm{e}}$re-Song-Strominger boundary conditions. The WCFT calculation amounts to the calculation of one-point functions on torus, whose high temperature limit can be approximated by using modular covariance of WCFT, similar to the derivation of Cardy formula. The bulk process is given by a tadpole diagram, with a massive spinning particle propagates from the infinity to the horizon, and splits into particle and antiparticle which annihilate after going around the horizon of BTZ black holes. The agreement between the bulk and WCFT calculations indicates that the black hole geometries in asymptotically AdS$_3$ spacetimes can emerge upon coarse-graining over microstates in WCFTs, similar to the results of Kraus and Maloney in the context of AdS/CFT[arXiv:1608.03284].

hep-th↗

arXiv2019

A General Construction and Encoder Implementation of Polar Codes

Similar to existing codes, puncturing and shortening are two general ways to obtain an arbitrary code length and code rate for polar codes. When some of the coded bits are punctured or shortened, it is equivalent to a situation in which the underlying channels of the polar codes are different. Therefore, the quality of bit channels with puncturing or shortening differ from the original qualities, which can greatly affect the construction of polar codes. In this paper, a general construction of polar codes is studied in two aspects: 1) the theoretical foundation of the construction; and 2) the hardware implementation of polar codes encoders. In contrast to the original identical and independent binary-input, memoryless, symmetric (BMS) channels, these underlying BMS channels can be different. For binary erasure channel (BEC) channels, recursive equations can be employed assuming independent BMS channels. For all other channel types, the proposed general construction of polar codes is based on the existing Tal-Vardy's procedure. The symmetric property and the degradation relationship are shown to be preserved under the general setting, rendering the possibility of a modification of Tal-Vardy's procedure. Simulation results clearly show improved error performance with re-ordering using the proposed new procedures. In terms of hardware, a novel pruned folded encoder architecture is proposed which saves the computation for the beginning frozen bits. Implementation results show the pruned encoder achieve 28% throughput improvement.

cs.IT↗

arXiv2019

Building a mixed-lingual neural TTS system with only monolingual data

When deploying a Chinese neural text-to-speech (TTS) synthesis system, one of the challenges is to synthesize Chinese utterances with English phrases or words embedded. This paper looks into the problem in the encoder-decoder framework when only monolingual data from a target speaker is available. Specifically, we view the problem from two aspects: speaker consistency within an utterance and naturalness. We start the investigation with an Average Voice Model which is built from multi-speaker monolingual data, i.e. Mandarin and English data. On the basis of that, we look into speaker embedding for speaker consistency within an utterance and phoneme embedding for naturalness and intelligibility and study the choice of data for model training. We report the findings and discuss the challenges to build a mixed-lingual TTS system with only monolingual data.

cs.CL↗

arXiv2019

An Experimental-based Review of Image Enhancement and Image Restoration Methods for Underwater Imaging

Underwater images play a key role in ocean exploration, but often suffer from severe quality degradation due to light absorption and scattering in water medium. Although major breakthroughs have been made recently in the general area of image enhancement and restoration, the applicability of new methods for improving the quality of underwater images has not specifically been captured. In this paper, we review the image enhancement and restoration methods that tackle typical underwater image impairments, including some extreme degradations and distortions. Firstly, we introduce the key causes of quality reduction in underwater images, in terms of the underwater image formation model (IFM). Then, we review underwater restoration methods, considering both the IFM-free and the IFM-based approaches. Next, we present an experimental-based comparative evaluation of state-of-the-art IFM-free and IFM-based methods, considering also the prior-based parameter estimation algorithms of the IFM-based methods, using both subjective and objective analysis (the used code is freely available at https://github.com/wangyanckxx/Single-Underwater-Image-Enhancement-and-Color-Restoration). Starting from this study, we pinpoint the key shortcomings of existing methods, drawing recommendations for future research in this area. Our review of underwater image enhancement and restoration provides researchers with the necessary background to appreciate challenges and opportunities in this important field.

eess.IV↗

arXiv2019

Enhancement of Underwater Images with Statistical Model of Background Light and Optimization of Transmission Map

Underwater images often have severe quality degradation and distortion due to light absorption and scattering in the water medium. A hazed image formation model is widely used to restore the image quality. It depends on two optical parameters: the background light and the transmission map. Underwater images can also be enhanced by color and contrast correction from the perspective of image processing. In this paper, we propose an effective underwater image enhancement method for underwater images in composition of underwater image restoration and color correction. Firstly, a manually annotated background lights (MABLs) database is developed. With reference to the relationship between MABLs and the histogram distributions of various underwater images, robust statistical models of BLs estimation are provided. Next, the TM of R channel is roughly estimated based on the new underwater dark channel prior via the statistic of clear and high resolution underwater images, then a scene depth map based on the underwater light attenuation prior and an adjusted reversed saturation map are applied to compensate and modify the coarse TM of R channel. Next, TMs of G-B channels are estimated based on the difference of attenuation ratios between R channel and G-B channels. Finally, to improve the color and contrast of the restored image with a natural appearance, a variation of white balance is introduced as post-processing. In order to guide the priority of underwater image enhancement, sufficient evaluations are conducted to discuss the impacts of the key parameters including BL and TM, and the importance of the color correction. Comparisons with other state-of-the-art methods demonstrate that our proposed underwater image enhancement method can achieve higher accuracy of estimated BLs, less computation time, more superior performance, and more valuable information retention.

eess.IV↗

arXiv2019

Femtosecond relativistic electron beam with reduced timing jitter from THz-driven beam compression

We propose and demonstrate a novel method to reduce the pulse width and timing jitter of a relativistic electron beam through THz-driven beam compression. In this method the longitudinal phase space of a relativistic electron beam is manipulated by a linearly polarized THz pulse in a dielectric tube such that the bunch tail has a higher velocity than the bunch head, which allows simultaneous reduction of both pulse width and timing jitter after passing through a drift. In this experiment, the beam is compressed by more than a factor of four from 130 fs to 28 fs with the arrival time jitter also reduced from 97 fs to 36 fs, opening up new opportunities in using pulsed electron beams for studies of ultrafast dynamics. This technique extends the well known rf buncher to the THz frequency and may have a strong impact in accelerator and ultrafast science facilities that require femtosecond electron beams with tight synchronization to external lasers.

physics.acc-ph↗

arXiv2019

Rényi Mutual Information in Holographic Warped CFTs

The study of Rényi mutual information (RMI) sheds light on the AdS/CFT correspondence beyond classical order. In this article, we study the Rényi mutual information between two intervals at large distance in two-dimensional holographic warped conformal field theory, which is conjectured to be dual to the gravity on AdS3 or warped AdS3 spacetime under the Dirichlet-Newman boundary conditions. By using the operator product expansion of twist operators up to level 3, we read the leading oder and the next-toleading order RMI in the large central charge and small cross-ratio limits. The leading order result is furthermore confirmed using the conformal block expansion. Finally, we match the next-to-leading result by a 1-loop calculation in the bulk.

hep-th↗

arXiv2019

Entanglement and chaos in warped conformal field theories

Various aspects of warped conformal field theories (WCFTs) are studied including entanglement entropy on excited states, the Renyi entropy after a local quench, and out-of-time-order four-point functions. Assuming a large central charge and dominance of the vacuum block in the conformal block expansion, (i) we calculate the single-interval entanglement entropy on an excited state, matching previous finite temperature results by changing the ensemble; and (ii) we show that WCFTs are maximally chaotic, a result that is compatible with the existence of black holes in the holographic duals. Finally, we relax the aforementioned assumptions and study the time evolution of the Renyi entropy after a local quench. We find that the change in the Renyi entropy is topological, vanishing at early and late times, and nonvanishing in between only for charged states in spectrally-flowed WCFTs.

hep-th↗

arXiv2019

Optimal Puncturing of Polar Codes With a Fixed Information Set

For a given polar code construction, the existing literature on puncturing for polar codes focuses in finding the optimal puncturing pattern, and then re-selecting the information set. This paper devotes itself to find the optimal puncturing pattern when the information set is fixed. Puncturing the coded bits corresponding to the worst quality bit channels, called the worst quality puncturing (WQP), is proposed, which is analyzed to minimize the bit channel quality loss at the punctured positions. Simulation results show that WQP outperforms the best existing puncturing schemes when the information set is fixed.

cs.IT↗

arXiv2019

Terahertz oscilloscope for recording time information of ultrashort electron beams

We propose and demonstrate a Terahertz (THz) oscilloscope for recording time information of an ultrashort electron beam. By injecting a laser-driven THz pulse with circular polarization into a dielectric tube, the electron beam is swept helically such that the time information is uniformly encoded into the angular distribution that allows one to characterize both the temporal profile and timing jitter of an electron beam. The dynamic range of the measurement in such a configuration is significantly increased compared to deflection with a linearly polarized THz pulse. With this THz oscilloscope, nearly 50-fold longitudinal compression of a relativistic electron beam to about 15 fs (rms) is directly visualized with its arrival time determined with 3 fs accuracy. This technique bridges the gap between streaking of photoelectrons with optical lasers and deflection of relativistic electron beams with radio-frequency deflectors, and should have wide applications in many ultrashort electron beam based facilities.

physics.acc-ph↗

arXiv2018

A note on the complex SYK model and warped CFTs

We discuss the connections between the complex SYK model at the conformal limit and warped conformal field theories. Both theories have an $SL(2,R) \times U(1)$ global symmetry. We present comparisons on symmetries, correlation functions, the effective action and the entropy formula. We also use modular covariance to reinterpret results in the complex SYK model.

hep-th↗

arXiv2018

Strings on warped AdS$_3$ via $T\bar{J}$ deformations

We study a toy model of the Kerr/CFT correspondence using string theory on AdS$_3 \times S^3$. We propose a single trace irrelevant deformation of the dual CFT generated by a vertex operator with spacetime dimensions (2,1). This operator shares the same quantum numbers as the integrable $T\bar{J}$ deformation of two-dimensional CFTs where $\bar{J}$ is a chiral $U(1)$ current. We show that the deformation is marginal on the worldsheet and that the target spacetime is deformed to null warped AdS$_3$ upon dimensional reduction. We also calculate the spectrum of the deformed theory on the cylinder and compare it to the field theory analysis of $T\bar{J}$-deformed CFTs.

hep-th↗