arXiv subjects

Xin Wang

Publications and source records attributed to Xin Wang.

At least 1,369 records · Page 76Linked to original sources

arXiv2021

Multi-Task Learning in Utterance-Level and Segmental-Level Spoof Detection

In this paper, we provide a series of multi-tasking benchmarks for simultaneously detecting spoofing at the segmental and utterance levels in the PartialSpoof database. First, we propose the SELCNN network, which inserts squeeze-and-excitation (SE) blocks into a light convolutional neural network (LCNN) to enhance the capacity of hidden feature selection. Then, we implement multi-task learning (MTL) frameworks with SELCNN followed by bidirectional long short-term memory (Bi-LSTM) as the basic model. We discuss MTL in PartialSpoof in terms of architecture (uni-branch/multi-branch) and training strategies (from-scratch/warm-up) step-by-step. Experiments show that the multi-task model performs relatively better than single-task models. Also, in MTL, a binary-branch architecture more adequately utilizes information from two levels than a uni-branch model. For the binary-branch architecture, fine-tuning a warm-up model works better than training from scratch. Models can handle both segment-level and utterance-level predictions simultaneously overall under a binary-branch multi-task architecture. Furthermore, the multi-task model trained by fine-tuning a segmental warm-up model performs relatively better at both levels except on the evaluation set for segmental detection. Segmental detection should be explored further.

cs.SD↗

arXiv2021

Fast binomial-code holonomic quantum computation with ultrastrong light-matter coupling

We propose a protocol for bosonic binomial-code nonadiabatic holonomic quantum computation in a system composed of an artificial atom ultrastrongly coupled to a cavity resonator. In our protocol, the binomial codes, formed by superpositions of Fock states, can greatly save physical resources to correct errors in quantum computation. We apply to the system strong driving fields designed by shortcuts-to-adiabatic methods. This reduces the gate time to tens of nanoseconds. Decoherence of the system accumulated over time can be effectively reduced by such a fast evolution. Noise induced by control imperfections can be suppressed by a systematic-error-sensitivity nullification method, thus the protocol is mostly insensitive to such noises. As a result, this protocol can rapidly ($\sim 35$ ns) generate fault-tolerant and high-fidelity ($\gtrsim 98\%$ with experimentally realistic parameters) quantum gates.

quant-ph↗

arXiv2021

A Hybrid Quantum-Classical Hamiltonian Learning Algorithm

Hamiltonian learning is crucial to the certification of quantum devices and quantum simulators. In this paper, we propose a hybrid quantum-classical Hamiltonian learning algorithm to find the coefficients of the Pauli operator components of the Hamiltonian. Its main subroutine is the practical log-partition function estimation algorithm, which is based on the minimization of the free energy of the system. Concretely, we devise a stochastic variational quantum eigensolver (SVQE) to diagonalize the Hamiltonians and then exploit the obtained eigenvalues to compute the free energy's global minimum using convex optimization. Our approach not only avoids the challenge of estimating von Neumann entropy in free energy minimization, but also reduces the quantum resources via importance sampling in Hamiltonian diagonalization, facilitating the implementation of our method on near-term quantum devices. Finally, we demonstrate our approach's validity by conducting numerical experiments with Hamiltonians of interest in quantum many-body physics.

quant-ph↗

arXiv2021

Robust Object Detection via Instance-Level Temporal Cycle Confusion

Building reliable object detectors that are robust to domain shifts, such as various changes in context, viewpoint, and object appearances, is critical for real-world applications. In this work, we study the effectiveness of auxiliary self-supervised tasks to improve the out-of-distribution generalization of object detectors. Inspired by the principle of maximum entropy, we introduce a novel self-supervised task, instance-level temporal cycle confusion (CycConf), which operates on the region features of the object detectors. For each object, the task is to find the most different object proposals in the adjacent frame in a video and then cycle back to itself for self-supervision. CycConf encourages the object detector to explore invariant structures across instances under various motions, which leads to improved model robustness in unseen domains at test time. We observe consistent out-of-domain performance improvements when training object detectors in tandem with self-supervised tasks on large-scale video datasets (BDD100K and Waymo open data). The joint training framework also establishes a new state-of-the-art on standard unsupervised domain adaptative detection benchmarks (Cityscapes, Foggy Cityscapes, and Sim10K). The code and models are available at https://github.com/xinw1012/cycle-confusion.

cs.CV↗

arXiv2021

Improved Lower Bounds for Secure Codes and Related Structures

Secure codes are widely-studied combinatorial structures which were introduced for traitor tracing in broadcast encryption. To determine the maximum size of such structures is the main research objective. In this paper, we investigate the lower bounds for secure codes and their related structures. First, we give some improved lower bounds for the rates of $2$-frameproof codes and $\overline{2}$-separable codes for slightly large alphabet size. Then we improve the lower bounds for the rate of some related structures, i.e., strongly $2$-separable matrices and $2$-cancellative set families. Finally, we give a general method to derive new lower bounds for strongly $t$-separable matrices and $t$-cancellative set families for $t\ge 3.$

cs.IT↗

arXiv2021

Symmetric distinguishability as a quantum resource

We develop a resource theory of symmetric distinguishability, the fundamental objects of which are elementary quantum information sources, i.e., sources that emit one of two possible quantum states with given prior probabilities. Such a source can be represented by a classical-quantum state of a composite system $XA$, corresponding to an ensemble of two quantum states, with $X$ being classical and $A$ being quantum. We study the resource theory for two different classes of free operations: $(i)$ ${\rm{CPTP}}_A$, which consists of quantum channels acting only on $A$, and $(ii)$ conditional doubly stochastic (CDS) maps acting on $XA$. We introduce the notion of symmetric distinguishability of an elementary source and prove that it is a monotone under both these classes of free operations. We study the tasks of distillation and dilution of symmetric distinguishability, both in the one-shot and asymptotic regimes. We prove that in the asymptotic regime, the optimal rate of converting one elementary source to another is equal to the ratio of their quantum Chernoff divergences, under both these classes of free operations. This imparts a new operational interpretation to the quantum Chernoff divergence. We also obtain interesting operational interpretations of the Thompson metric, in the context of the dilution of symmetric distinguishability.

quant-ph↗

arXiv2021

Intra-Day Variability Observations of Two Dozens of Blazars at 4.8 GHz

Two dozens of radio loud active galactic nuclei (AGNs) have been observed with Urumqi 25 m radio telescope in order to search for intra-day variability (IDV). The target sources are blazars (namely flat spectrum radio quasars and BL Lac objects) which are mostly selected from the observing list of RadioAstron AGN monitoring campaigns. The observations were carried out at 4.8 GHz in two sessions of 8-12 February 2014 and 7-9 March respectively. We report the data reduction and the first results of observations. The results show that the majority of the blazars exhibit IDV in 99.9% confidence level, some of them show quite strong IDV. We find the strong IDV of blazar 1357 + 769 for the first time. The IDV at centimeter-wavelength is believed to be predominately caused by the scintillation of blazar emission through the local interstellar medium in a few hundreds parsecs away from Sun. No significant correlation between the IDV strength and either redshift or Galactic latitude is found in our sample. The IDV timescale along with source structure and brightness temperature analysis will be presented in a forthcoming paper.

astro-ph.HE↗

arXiv2021

Interpreting Attributions and Interactions of Adversarial Attacks

This paper aims to explain adversarial attacks in terms of how adversarial perturbations contribute to the attacking task. We estimate attributions of different image regions to the decrease of the attacking cost based on the Shapley value. We define and quantify interactions among adversarial perturbation pixels, and decompose the entire perturbation map into relatively independent perturbation components. The decomposition of the perturbation map shows that adversarially-trained DNNs have more perturbation components in the foreground than normally-trained DNNs. Moreover, compared to the normally-trained DNN, the adversarially-trained DNN have more components which mainly decrease the score of the true category. Above analyses provide new insights into the understanding of adversarial attacks.

cs.LG↗

arXiv2021

Distilling Holistic Knowledge with Graph Neural Networks

Knowledge Distillation (KD) aims at transferring knowledge from a larger well-optimized teacher network to a smaller learnable student network.Existing KD methods have mainly considered two types of knowledge, namely the individual knowledge and the relational knowledge. However, these two types of knowledge are usually modeled independently while the inherent correlations between them are largely ignored. It is critical for sufficient student network learning to integrate both individual knowledge and relational knowledge while reserving their inherent correlation. In this paper, we propose to distill the novel holistic knowledge based on an attributed graph constructed among instances. The holistic knowledge is represented as a unified graph-based embedding by aggregating individual knowledge from relational neighborhood samples with graph neural networks, the student network is learned by distilling the holistic knowledge in a contrastive manner. Extensive experiments and ablation studies are conducted on benchmark datasets, the results demonstrate the effectiveness of the proposed method. The code has been published in https://github.com/wyc-ruiker/HKD

cs.CV↗

arXiv2021

Transverse mode selective laser with gain regulation by a digital micromirror device

A transverse mode selective laser system with gain regulation by a digital micromirror device (DMD) is presented in this letter. The gain regulation in laser medium is adjusted by the switch of the patterns loaded on DMD. Structured pump beam patterns can be obtained after the reflection of the loaded patterns on DMD, and then it's defocused into a microchip laser medium by a short focal lens, so that the pump patterns can be transferred to the gain medium to regulate the gain distribution. Corresponding structured laser beams can be generated by this laser system. The laser beam pattern can be regulated easily and quickly, by switching the loaded patterns on DMD. Through this method, we show a simple and flexible laser system to generate on-demand laser beam patterns.

physics.optics↗

arXiv2021

Unsupervised Domain Adaptation for Retinal Vessel Segmentation with Adversarial Learning and Transfer Normalization

Retinal vessel segmentation plays a key role in computer-aided screening, diagnosis, and treatment of various cardiovascular and ophthalmic diseases. Recently, deep learning-based retinal vessel segmentation algorithms have achieved remarkable performance. However, due to the domain shift problem, the performance of these algorithms often degrades when they are applied to new data that is different from the training data. Manually labeling new data for each test domain is often a time-consuming and laborious task. In this work, we explore unsupervised domain adaptation in retinal vessel segmentation by using entropy-based adversarial learning and transfer normalization layer to train a segmentation network, which generalizes well across domains and requires no annotation of the target domain. Specifically, first, an entropy-based adversarial learning strategy is developed to reduce the distribution discrepancy between the source and target domains while also achieving the objective of entropy minimization on the target domain. In addition, a new transfer normalization layer is proposed to further boost the transferability of the deep network. It normalizes the features of each domain separately to compensate for the domain distribution gap. Besides, it also adaptively selects those feature channels that are more transferable between domains, thus further enhancing the generalization performance of the network. We conducted extensive experiments on three regular fundus image datasets and an ultra-widefield fundus image dataset, and the results show that our approach yields significant performance gains compared to other state-of-the-art methods.

eess.IV↗

arXiv2021

Dynamic Multi-scale Convolution for Dialect Identification

Time Delay Neural Networks (TDNN)-based methods are widely used in dialect identification. However, in previous work with TDNN application, subtle variant is being neglected in different feature scales. To address this issue, we propose a new architecture, named dynamic multi-scale convolution, which consists of dynamic kernel convolution, local multi-scale learning, and global multi-scale pooling. Dynamic kernel convolution captures features between short-term and long-term context adaptively. Local multi-scale learning, which represents multi-scale features at a granular level, is able to increase the range of receptive fields for convolution operation. Besides, global multi-scale pooling is applied to aggregate features from different bottleneck layers in order to collect information from multiple aspects. The proposed architecture significantly outperforms state-of-the-art system on the AP20-OLR-dialect-task of oriental language recognition (OLR) challenge 2020, with the best average cost performance (Cavg) of 0.067 and the best equal error rate (EER) of 6.52%. Compared with the known best results, our method achieves 9% of Cavg and 45% of EER relative improvement, respectively. Furthermore, the parameters of proposed model are 91% fewer than the best known model.

cs.CL↗

arXiv2021

An adaptive high-order surface finite element method for the self-consistent field theory on general curved surfaces

In this paper, we develop an adaptive high-order surface finite element method (FEM) incorporating the spectral deferred correction method for chain contour discretization to solve polymeric self-consistent field equations on general curved surfaces. The high-order surface FEM is obtained by the high-order surface geometrical approximation and the high-order function space approximation. Numerical results demonstrate that the precision order of these methods is consistent with the theoretical prediction. In order to describe the sharp interface in the strongly segregated system more accurately, an adaptive FEM equipped with a new Log marking strategy is proposed. Compared with the traditional strategy, the Log marking strategy can not only label the elements that need to be refined or coarsened, but also give the refined or coarsened times, which can make full use of the information of a posterior error estimator and improve the ecciency of the adaptive algorithm. To demonstrate the power of our approach, we investigate the self-assembled patterns of diblock copolymers on several distinct curved surfaces. Numerical results illustrate the ecciency of the proposed method, especially for strongly segregated systems with economical discretization nodes.

math.NA↗

arXiv2021

T$_k$ML-AP: Adversarial Attacks to Top-$k$ Multi-Label Learning

Top-$k$ multi-label learning, which returns the top-$k$ predicted labels from an input, has many practical applications such as image annotation, document analysis, and web search engine. However, the vulnerabilities of such algorithms with regards to dedicated adversarial perturbation attacks have not been extensively studied previously. In this work, we develop methods to create adversarial perturbations that can be used to attack top-$k$ multi-label learning-based image annotation systems (TkML-AP). Our methods explicitly consider the top-$k$ ranking relation and are based on novel loss functions. Experimental evaluations on large-scale benchmark datasets including PASCAL VOC and MS COCO demonstrate the effectiveness of our methods in reducing the performance of state-of-the-art top-$k$ multi-label learning methods, under both untargeted and targeted attacks.

cs.CV↗

arXiv2021

A Hypothesis for the Aesthetic Appreciation in Neural Networks

This paper proposes a hypothesis for the aesthetic appreciation that aesthetic images make a neural network strengthen salient concepts and discard inessential concepts. In order to verify this hypothesis, we use multi-variate interactions to represent salient concepts and inessential concepts contained in images. Furthermore, we design a set of operations to revise images towards more beautiful ones. In experiments, we find that the revised images are more aesthetic than the original ones to some extent.

cs.LG↗

arXiv2021

The reducing role of hydrogen peroxide on the formation of gold nanostructures in aqueous microdroplets with dissolved tetrachloroaurate ions

A recent article by Lee et al. in Nature Communications has reported an intriguing phenomenon that gold nanostructures, AuNSs, can be spontaneously formed in aqueous microdroplets with dissolved tetrachloroaurate ions. The authors suggested three possible electron donors for the reduction of AuCl4-, including the strong electric field at the microdroplet surface, the hydroxyl ions, and the AuCl4- itself. However, we find that the hydrogen peroxide spontaneously produced at the microdroplets might be also responsible for the AuCl4- reduction.

physics.chem-ph↗

arXiv2021

Mode structure symmetry breaking of reversed shear Alfvén eigenmodes and its impact on the generation of parallel velocity asymmetries in energetic particle distribution

In this work, the gyrokinetic eigenvalue code LIGKA, the drift-kinetic/MHD hybrid code HMGC and the gyrokinetic full-f code TRIMEG-GKX are employed to study the mode structure details of Reversed Shear Alfvén Eigenmodes (RSAEs). Using the parameters from an ASDEX-Upgrade plasma, a benchmark with the three different physical models for RSAE without and with Energetic Particles (EPs) is carried out. Reasonable agreement has been found for the mode frequency and the growth rate. Mode structure symmetry breaking (MSSB) is observed when EPs are included, due to the EPs' non-perturbative effects. It is found that the MSSB properties are featured by a finite radial wave phase velocity, and the linear mode structure can be well described by an analytical complex Gaussian expression $A(s)=e^{- σ(s-s_0)^2}$ with complex parameters $σ$ and $s_0$, where $s$ is the normalized radial coordinate. The mode structure is distorted in opposite {manners} when the EP drive shifted from one side of $q_{min}$ to the other side, and specifically, a non-zero average radial wave number $\langle k_s\rangle$ with opposite signs is generated. The initial EP density profiles and the corresponding mode structures have been used as the input of HAGIS code to study the EP transport. The parallel velocity of EPs is generated in opposite directions, due to different values of the average radial wave number $\langle k_s\rangle$, corresponding to different initial EP density profiles with EP drive shifted away from the $q_{min}$.

physics.plasm-ph↗

arXiv2021

MOSNet: Deep Learning based Objective Assessment for Voice Conversion

Existing objective evaluation metrics for voice conversion (VC) are not always correlated with human perception. Therefore, training VC models with such criteria may not effectively improve naturalness and similarity of converted speech. In this paper, we propose deep learning-based assessment models to predict human ratings of converted speech. We adopt the convolutional and recurrent neural network models to build a mean opinion score (MOS) predictor, termed as MOSNet. The proposed models are tested on large-scale listening test results of the Voice Conversion Challenge (VCC) 2018. Experimental results show that the predicted scores of the proposed MOSNet are highly correlated with human MOS ratings at the system level while being fairly correlated with human MOS ratings at the utterance level. Meanwhile, we have modified MOSNet to predict the similarity scores, and the preliminary results show that the predicted scores are also fairly correlated with human ratings. These results confirm that the proposed models could be used as a computational evaluator to measure the MOS of VC systems to reduce the need for expensive human rating.

cs.SD↗