arXiv subjects

Xia Zhang

Publications and source records attributed to Xia Zhang.

At least 37 records · Page 2Linked to original sources

arXiv2023

Streaming Parrotron for on-device speech-to-speech conversion

We present a fully on-device streaming Speech2Speech conversion model that normalizes a given input speech directly to synthesized output speech. Deploying such a model on mobile devices pose significant challenges in terms of memory footprint and computation requirements. We present a streaming-based approach to produce an acceptable delay, with minimal loss in speech conversion quality, when compared to a reference state of the art non-streaming approach. Our method consists of first streaming the encoder in real time while the speaker is speaking. Then, as soon as the speaker stops speaking, we run the spectrogram decoder in streaming mode along the side of a streaming vocoder to generate output speech. To achieve an acceptable delay-quality trade-off, we propose a novel hybrid approach for look-ahead in the encoder which combines a look-ahead feature stacker with a look-ahead self-attention. We show that our streaming approach is almost 2x faster than real time on the Pixel4 CPU.

eess.AS↗

arXiv2023

Chemical models of adenine precursors cyanamide and carbodiimide in the interstellar medium

Cyanamide (NH2CN) and its isomer, carbodiimide (HNCNH), may form adenine in the interstellar medium (ISM) via a series of reactions. Therefore, they are considered key prebiotic molecules in the study of the origin of life. We used the three-phase NAUTILUS chemical code, which includes the gas, the dust surface, and the icy mantle, to investigate the formation and destruction of cyanamide and carbodiimide. We added over 200 new chemical reactions of the two isomers and related species, and established a relatively complete network. We applied cold core, hot corino/core and shock models to simulate the different physical environments, and found that the two isomers are mainly produced by the free radical reactions on grain surfaces. Our simulated results suggest that cyanamide and carbodiimide molecules come from surface chemistry at early evolutionary stages. Then they are released back to the gas phase, either by thermal process (in hot cores, hot corinos) or shock-induced desorption (in shock regions).We speculate that it is an inefficient route to form a tautomer of adenine by starting from molecules cyanoacetylene (C3NH), cyanamide and carbodiimide in ISM.

astro-ph.GA↗

arXiv2023

On monogamy and polygamy relations of multipartite systems

We study the monogamy and polygamy relations related to quantum correlations for multipartite quantum systems in a unified manner. It is known that any bipartite measure obeys monogamy and polygamy relations for the $r$-power of the measure. We show in a uniformed manner that the generalized monogamy and polygamy relations are transitive to other powers of the measure in weighted forms. We demonstrate that our weighted monogamy and polygamy relations are stronger than recently available relations. Comparisons are given in detailed examples which show that our results are stronger in both situations.

quant-ph↗

arXiv2022

Guiding Neural Entity Alignment with Compatibility

Entity Alignment (EA) aims to find equivalent entities between two Knowledge Graphs (KGs). While numerous neural EA models have been devised, they are mainly learned using labelled data only. In this work, we argue that different entities within one KG should have compatible counterparts in the other KG due to the potential dependencies among the entities. Making compatible predictions thus should be one of the goals of training an EA model along with fitting the labelled data: this aspect however is neglected in current methods. To power neural EA models with compatibility, we devise a training framework by addressing three problems: (1) how to measure the compatibility of an EA model; (2) how to inject the property of being compatible into an EA model; (3) how to optimise parameters of the compatibility model. Extensive experiments on widely-used datasets demonstrate the advantages of integrating compatibility within EA models. In fact, state-of-the-art neural EA models trained within our framework using just 5\% of the labelled data can achieve comparable effectiveness with supervised training using 20\% of the labelled data.

cs.CL↗

arXiv2022

A Scalable Model Specialization Framework for Training and Inference using Submodels and its Application to Speech Model Personalization

Model fine-tuning and adaptation have become a common approach for model specialization for downstream tasks or domains. Fine-tuning the entire model or a subset of the parameters using light-weight adaptation has shown considerable success across different specialization tasks. Fine-tuning a model for a large number of domains typically requires starting a new training job for every domain posing scaling limitations. Once these models are trained, deploying them also poses significant scalability challenges for inference for real-time applications. In this paper, building upon prior light-weight adaptation techniques, we propose a modular framework that enables us to substantially improve scalability for model training and inference. We introduce Submodels that can be quickly and dynamically loaded for on-the-fly inference. We also propose multiple approaches for training those Submodels in parallel using an embedding space in the same training job. We test our framework on an extreme use-case which is speech model personalization for atypical speech, requiring a Submodel for each user. We obtain 128x Submodel throughput with a fixed computation budget without a loss of accuracy. We also show that learning a speaker-embedding space can scale further and reduce the amount of personalization training data required per speaker.

eess.AS↗

arXiv2022

Reflectors to quantales

In this paper, we show that marked quantales have a reflection into quantales. To obtain the reflection we construct free quantales over marked quantales using appropriate lower sets. A marked quantale is a posemigroup in which certain admissible subsets are required to have joins, and multiplication distributes over these. Sometimes are the admissible subsets in question specified by means of a so-called selection function. A distinguishing feature of the study of marked quantales is that a small collection of axioms of an elementary nature allows one to do much that is traditional at the level of quantales. The axioms are sufficiently general to include as examples of marked quantales the classes of posemigroups, $σ$-quantales, prequantales and quantales. Furthermore, we discuss another reflection to quantales obtained by the injective hull of a posemigroup.

math.CT↗

arXiv2022

High-quality Task Division for Large-scale Entity Alignment

Entity Alignment (EA) aims to match equivalent entities that refer to the same real-world objects and is a key step for Knowledge Graph (KG) fusion. Most neural EA models cannot be applied to large-scale real-life KGs due to their excessive consumption of GPU memory and time. One promising solution is to divide a large EA task into several subtasks such that each subtask only needs to match two small subgraphs of the original KGs. However, it is challenging to divide the EA task without losing effectiveness. Existing methods display low coverage of potential mappings, insufficient evidence in context graphs, and largely differing subtask sizes. In this work, we design the DivEA framework for large-scale EA with high-quality task division. To include in the EA subtasks a high proportion of the potential mappings originally present in the large EA task, we devise a counterpart discovery method that exploits the locality principle of the EA task and the power of trained EA models. Unique to our counterpart discovery method is the explicit modelling of the chance of a potential mapping. We also introduce an evidence passing mechanism to quantify the informativeness of context entities and find the most informative context graphs with flexible control of the subtask size. Extensive experiments show that DivEA achieves higher EA performance than alternative state-of-the-art solutions.

cs.AI↗

arXiv2022

Polaritonic Critical Coupling in a Hybrid Quasi-Bound States in the Continuum Cavity-WS$_2$ Monolayer System

We theoretically propose and numerically demonstrate that perfect feeding of a polaritonic system with full electromagnetic energy under one-port beam incidence, referred to as polaritonic critical coupling, can be achieved in a hybrid dielectric metasurface-WS$_2$ monolayer structure. Polaritonic critical coupling, where the critical coupling and strong coupling are simultaneously attained, is determined by the relative damping rates of the cavity resonance, $\rm γ_Q$, provided by a symmetry-protected quasi-bound states in the continuum, and excitonic resonance of WS$_2$ monolayer, $\rm γ_X$. We reveal that the population of the polariton states can be tuned by the asymmetric parameter of the quasi-bound states in the continuum. Furthermore, polaritonic critical coupling is achieved in the designed system while $\rm γ_Q=γ_X$ and only strong coupling is achieved while $\rm γ_Q\neqγ_X$. This work enriches the study of polaritonic physics with controlled absorbance and may guide the design and application of efficient polariton-based light-emitting or lasing devices.

physics.optics↗

arXiv2022

Absorbance Enhancement of Monolayer MoS$_2$ in a Perfect Absorbing System

We reveal numerically and experimentally that dielectric resonance can enhance the absorbance and emission of monolayer MoS$_2$. By quantifying the absorbance of the Si disk resonators and the monolayer MoS$_2$ separately, a model taking into account of absorbance as well as quantum efficiency modifications by the dielectric disk resonators successfully explains the observed emission enhancement under the normal light incidence. It is demonstrated that the experimentally observed emission enhancement at different pump wavelength results from the absorbance enhancement, which compensates the emission quenching by the disk resonators. In order to further maximize the absorbance value of monolayer MoS$_2$, a perfect absorbing structure is proposed. By placing a Au mirror beneath the Si nanodisks, the incident electromagnetic power is fully absorbed by the hybrid monolayer MoS$_2$-disk system. It is demonstrated that the electromagnetic power is re-distributed within the hybrid structure and 53\% of the total power is absorbed by the monolayer MoS$_2$ at the perfect absorbing wavelength.

physics.optics↗

arXiv2022

Radio Transient Detection with Closure Products and Machine Learning

For transient sources with timescales of 1-100 seconds, standardized imaging for all observations at each time step become impossible as large modern interferometers produce significantly large data volumes in this observation time frame. Here we propose a method based on machine learning and using interferometric closure products as input features to detect transient source candidates directly from the spatial frequency domain without imaging. We train a simple neural network classifier on a synthetic dataset of Noise/Transient/RFI events, which we construct to tackle the lack of labelled data. We also use the hyper-parameter dropout rate of the model to allow the model to approximate Bayesian inference, and select the optimal dropout rate to match the posterior prediction to the actual underlying probability distribution of the detected events. The overall F1-score of the classifier on the simulated dataset is greater than 85\%, with the signal-to-noise at 7$σ$. The performance of the trained neural network with Monte Carlo dropout is evaluated on semi-real data, which includes a simulated transient source and real noise. This classifier accurately identifies the presence of transient signals in the detectable signal-to-noise levels (above 4$σ$) with the optimal variance. Our findings suggest that a feasible radio transient classifier can be built up with only simulated data for applying to the prediction of real observation, even in the absence of annotated real samples for the purpose of training.

astro-ph.IM↗

arXiv2022

Trotter-Kato Approximations of Impulsive Neutral SPDEs in Hilbert Spaces

This paper studies a class of impulsive neutral stochastic partial differential equations in real Hilbert spaces. The main goal here is to consider the Trotter-Kato approximations of mild solutions of such equations in the $p$th-mean ($p\geq2$). As an application, a classical limit theorem on the dependence of such equations on a parameter is obtained. The novelty of this paper is that the combination of this approximating system and such equations has not been considered before.

math.PR↗

arXiv2022

Effect of surface H$_2$ on molecular hydrogen formation on interstellar grains

We investigate how the existence of hydrogen molecules on grain surfaces may affect H$_2$ formation efficiency in diffuse and translucent clouds. Hydrogen molecules are able to reduce the desorption energy of H atoms on grain surfaces in models. The detailed microscopic Monte Carlo method is used to perform model simulations. We found that the impact of the existence of H$_2$ on H$_2$ formation efficiency strongly depends on the diffusion barriers of H$_2$ on grain surfaces. Diffuse cloud models that do not consider surface H$_2$ predict that H atom recombination efficiency is above 0.5 over a grain temperature (T) range 10 K and 14 K. The adopted H$_2$ diffusion barriers in diffuse cloud models that consider surface H$_2$ are 80$\%$ H$_2$ desorption energies so that H$_2$ can be trapped in stronger binding sites. Depending on model parameters, these diffuse cloud models predict that the recombination efficiency is between nearly 0 and 0.5 at 10 K $\leq$ T $\leq$ 14 K. Translucent cloud model results show that H$_2$ formation efficiency is not affected by the existence of surface H$_2$ if the adopted average H$_2$ diffusion barrier on grain surfaces is low (194 K) so that H$_2$ can diffuse rapidly on grain surfaces. However, the recombination efficiency can drop to below 0.002 at T $\geq$ 10 K if higher average H$_2$ diffusion barrier is used (255 K) in translucent cloud models.

astro-ph.GA↗

arXiv2022

Chemical variations across the TMC-1 boundary: molecular tracers from translucent phase to dense phase

We investigated the chemical evolutions of gas phase and grain surface species across the Taurus molecular cloud-1 (TMC-1) filament from translucent phase to dense phase. By comparing observations with modeling results from an up-to-date chemical network, we examined the conversion processes for the carbon-, oxygen-, nitrogen- and sulfur-bearing species, i.e.from their initial atomic form to their main molecular reservoir form both in the gas phase and on the grain surface. The conversion processes were found to depend on the species and A$_V$. The effect of initial carbon to oxygen elemental abundances ratio (C/O) by varying O on the chemistry was explored, and an initial carbon elemental abundance of 2.5 $\times$ 10$^{-4}$ and a C/O ratio of 0.5 could best reproduce the abundances of most observed molecules at TMC-1 CP, where more than 90 molecules have been identified. Based on the TMC-1 condition, we predicted a varied grain ice composition during the evolutions of molecular clouds, with H$_2$O ice as the dominant ice composition at A$_V$ $>$ 4 mag, CO$_2$ ice as the dominant ice composition at A$_V$ $<$ 4 mag, while CO ice severely decreased at A$_V$ around 4--5 mag.

astro-ph.GA↗

arXiv2022

R-Matrix presentation of quantum affine algebra in type $A_{2n-1}^{(2)}$

In this paper, we give an RTT presentation of the twisted quantum affine algebra of type $A_{2n-1}^{(2)}$ and show that it is isomorphic to the Drinfeld new realization via the Gauss decomposition of the L-operators. This provides the first such presentation for twisted quantum affine algebras with nontrivial central element.

math.QA↗

arXiv2021

Ultra-Spatiotemporal Light Confinement in Dielectric Nanocavity Metasurfaces

Light concentration with strong temporal and spatial confinement is crucial for tailoring light-matter interaction. Electromagnetic cavity modes in photonic and plasmonic resonators provide platforms for optical field localization. Here, we propose a concept of quasi-bound states in the continuum gap cavity and reveal that ultra spatiotemporal confinements in free-space can be realized in a dielectric nanocavity metasurface. By introducing an asymmetric air slot in a nanodisk resonator, an ultra-high quality factor $\rm Q \sim 10^6$, accompanying an ultra-small effective mode volume, $\rm V_m \sim 10^{-2}$ $(λ/n)^3$ are achieved resulting in a Purcell factor of $\rm 10^6 (λ/n)^{-3}$ in the visible wavelength range. The toroidal dipole drives the electric and magnetic field concentration in the air gap with a generated vortex polarizing electric field. As an alternative to plasmonic and photonic crystal cavities, our study provides a more intriguing platform for engineering light-matter interaction to advance a plethora of fundamental studies and device applications, such as Purcell factor enhancement, room temperature strong coupling and nonlinear nanophotoncis.

physics.optics↗

arXiv2021

Trotter-Kato approximations of semilinear stochastic evolution equations in Hilbert spaces

Motivated by the work of T.E. Govindan in [5,8,9], this paper is concerned with a more general semilinear stochastic evolution equation. The difference between the equations considered in this paper and the previous one is that it makes some changes to the nonlinear function in random integral, which also depends on the probability distribution of stochastic process at that time. First, this paper considers the existence and uniqueness of mild solutions for such equations. Furthermore, Trotter-Kato approximation system is introduced for the mild solutions, and the weak convergence of induced probability measures and zeroth-order approximations are obtained. Then we consider the classical limit theorem about the parameter dependence of this kind of equations. Finally, an example of stochastic partial differential equation is given to illustrate our results.

math.PR↗

arXiv2021

Wide-Angle Invisible Dielectric Metasurface Driven by Transverse Kerker Scattering

Interference is the cornerstone of Huygens source design for reshaping and controlling scattering patterns. The conventional underpinning principle, such as for the Kerker effect, is the interference of electric and magnetic dipole and quadrupole modes. Here a route to realize transverse Kerker scattering through employing only the interference between the electric dipole and magnetic quadrupole is demonstrated. The proposed approach is numerically validated in an ultra-thin Silicon square nanoplate metasurface, and is further verified by multipole decomposition. The metasurface is shown to be invisible fornear-infrared wavelengths and with an enhanced electric field in the region of the nanoparticle. Additionally, we develop further the proposed approach with practical implementation for invisibility applications by exploring the effects of the aspect ratio of the square plate nanoresonator, the inter-particle separation, and the presence of a substrate. Further it is demonstrated that invisibility can be observed at oblique incidence up to 60° for a transverse magnetic plane wave. The results are relevant for Huygens metasurface design for perfect reflectors, invisibility and devices for harmonic generation manipulation.

physics.optics↗

arXiv2020

Constructive and Destructive Interference of Kerker-type Scattering in an Ultra-thin Silicon Huygens Metasurface

High refractive index dielectric nanoparticles have provided a new platform for exotic light manipulation through the interference of multipole modes. The Kerker effect is one example of a Huygens source design. Rather than exploiting interference between the electric dipole and magnetic dipole, as in many conventional Huygens source designs, we explore Kerker-type suppressed backward scattering mediated by the dominant electric dipole, toroidal dipole and magnetic quadrupole. These modes are provided by a designed and fabricated CMOS compatible ultra-thin Silicon nanodisk metasurface with a suppressed magnetic dipole contribution, and verified through multipole decomposition. The non-trivial substrate effect is considered using a semi-analytical transfer matrix model. The model successfully predicts the observed reflection dip. By applying a general criterion for constructive and destructive interference, it is shown that while constructive interference occurs between the electric and toroidal dipole contributions, the experimentally observed suppressed backward Kerker-type scattering arises from the destructive interference between backward scattered contributions due to the total electric dipole and the magnetic quadrupole. Our study paves the way towards new types of Huygens sources or metasurface design, such as for peculiar transverse Kerker scattering.

physics.optics↗