arXiv subjects

Xu Li

Publications and source records attributed to Xu Li.

At least 127 records · Page 7Linked to original sources

arXiv2022

DeepSTI: Towards Tensor Reconstruction using Fewer Orientations in Susceptibility Tensor Imaging

Susceptibility tensor imaging (STI) is an emerging magnetic resonance imaging technique that characterizes the anisotropic tissue magnetic susceptibility with a second-order tensor model. STI has the potential to provide information for both the reconstruction of white matter fiber pathways and detection of myelin changes in the brain at mm resolution or less, which would be of great value for understanding brain structure and function in healthy and diseased brain. However, the application of STI in vivo has been hindered by its cumbersome and time-consuming acquisition requirement of measuring susceptibility induced MR phase changes at multiple (usually more than six) head orientations. This complexity is enhanced by the limitation in head rotation angles due to physical constraints of the head coil. As a result, STI has not yet been widely applied in human studies in vivo. In this work, we tackle these issues by proposing an image reconstruction algorithm for STI that leverages data-driven priors. Our method, called DeepSTI, learns the data prior implicitly via a deep neural network that approximates the proximal operator of a regularizer function for STI. The dipole inversion problem is then solved iteratively using the learned proximal network. Experimental results using both simulation and in vivo human data demonstrate great improvement over state-of-the-art algorithms in terms of the reconstructed tensor image, principal eigenvector maps and tractography results, while allowing for tensor reconstruction with MR phase measured at much less than six different orientations. Notably, promising reconstruction results are achieved by our method from only one orientation in human in vivo, and we demonstrate a potential application of this technique for estimating lesion susceptibility anisotropy in patients with multiple sclerosis.

eess.IV↗

arXiv2022

A Hierarchical Speaker Representation Framework for One-shot Singing Voice Conversion

Typically, singing voice conversion (SVC) depends on an embedding vector, extracted from either a speaker lookup table (LUT) or a speaker recognition network (SRN), to model speaker identity. However, singing contains more expressive speaker characteristics than conversational speech. It is suspected that a single embedding vector may only capture averaged and coarse-grained speaker characteristics, which is insufficient for the SVC task. To this end, this work proposes a novel hierarchical speaker representation framework for SVC, which can capture fine-grained speaker characteristics at different granularity. It consists of an up-sampling stream and three down-sampling streams. The up-sampling stream transforms the linguistic features into audio samples, while one down-sampling stream of the three operates in the reverse direction. It is expected that the temporal statistics of each down-sampling block can represent speaker characteristics at different granularity, which will be engaged in the up-sampling blocks to enhance the speaker modeling. Experiment results verify that the proposed method outperforms both the LUT and SRN based SVC systems. Moreover, the proposed system supports the one-shot SVC with only a few seconds of reference audio.

eess.AS↗

arXiv2022

CGAR: Critic Guided Action Redistribution in Reinforcement Leaning

Training a game-playing reinforcement learning agent requires multiple interactions with the environment. Ignorant random exploration may cause a waste of time and resources. It's essential to alleviate such waste. As discussed in this paper, under the settings of the off-policy actor critic algorithms, we demonstrate that the critic can bring more expected discounted rewards than or at least equal to the actor. Thus, the Q value predicted by the critic is a better signal to redistribute the action originally sampled from the policy distribution predicted by the actor. This paper introduces the novel Critic Guided Action Redistribution (CGAR) algorithm and tests it on the OpenAI MuJoCo tasks. The experimental results demonstrate that our method improves the sample efficiency and achieves state-of-the-art performance. Our code can be found at https://github.com/tairanhuang/CGAR.

cs.LG↗

arXiv2022

Inf-sup stabilized Scott--Vogelius pairs on general simplicial grids by Raviart--Thomas enrichment

This paper considers the discretization of the Stokes equations with Scott--Vogelius pairs of finite element spaces on arbitrary shape-regular simplicial grids. A novel way of stabilizing these pairs with respect to the discrete inf-sup condition is proposed and analyzed. The key idea consists in enriching the continuous polynomials of order $k$ of the Scott--Vogelius velocity space with appropriately chosen and explicitly given Raviart--Thomas bubbles. This approach is inspired by [Li/Rui, IMA J. Numer. Anal, 2021], where the case $k=1$ was studied. The proposed method is pressure-robust, with optimally converging $\boldsymbol{H}^1$-conforming velocity and a small $\boldsymbol{H}(\mathrm{div})$-conforming correction rendering the full velocity divergence-free. For $k\ge d$, with $d$ being the dimension, the method is parameter-free. Furthermore, it is shown that the additional degrees of freedom for the Raviart--Thomas enrichment and also all non-constant pressure degrees of freedom can be condensated, effectively leading to a pressure-robust, inf-sup stable, optimally convergent $\boldsymbol{P}_k \times P_0$ scheme. Aspects of the implementation are discussed and numerical studies confirm the analytic results.

math.NA↗

arXiv2022

Joint learning of object graph and relation graph for visual question answering

Modeling visual question answering(VQA) through scene graphs can significantly improve the reasoning accuracy and interpretability. However, existing models answer poorly for complex reasoning questions with attributes or relations, which causes false attribute selection or missing relation in Figure 1(a). It is because these models cannot balance all kinds of information in scene graphs, neglecting relation and attribute information. In this paper, we introduce a novel Dual Message-passing enhanced Graph Neural Network (DM-GNN), which can obtain a balanced representation by properly encoding multi-scale scene graph information. Specifically, we (i)transform the scene graph into two graphs with diversified focuses on objects and relations; Then we design a dual structure to encode them, which increases the weights from relations (ii)fuse the encoder output with attribute features, which increases the weights from attributes; (iii)propose a message-passing mechanism to enhance the information transfer between objects, relations and attributes. We conduct extensive experiments on datasets including GQA, VG, motif-VG and achieve new state of the art.

cs.CV↗

arXiv2022

Characterizing the adversarial vulnerability of speech self-supervised learning

A leaderboard named Speech processing Universal PERformance Benchmark (SUPERB), which aims at benchmarking the performance of a shared self-supervised learning (SSL) speech model across various downstream speech tasks with minimal modification of architectures and small amount of data, has fueled the research for speech representation learning. The SUPERB demonstrates speech SSL upstream models improve the performance of various downstream tasks through just minimal adaptation. As the paradigm of the self-supervised learning upstream model followed by downstream tasks arouses more attention in the speech community, characterizing the adversarial robustness of such paradigm is of high priority. In this paper, we make the first attempt to investigate the adversarial vulnerability of such paradigm under the attacks from both zero-knowledge adversaries and limited-knowledge adversaries. The experimental results illustrate that the paradigm proposed by SUPERB is seriously vulnerable to limited-knowledge adversaries, and the attacks generated by zero-knowledge adversaries are with transferability. The XAB test verifies the imperceptibility of crafted adversarial attacks.

cs.SD↗

arXiv2022

Spoofing-Aware Speaker Verification by Multi-Level Fusion

Recently, many novel techniques have been introduced to deal with spoofing attacks, and achieve promising countermeasure (CM) performances. However, these works only take the stand-alone CM models into account. Nowadays, a spoofing aware speaker verification (SASV) challenge which aims to facilitate the research of integrated CM and ASV models, arguing that jointly optimizing CM and ASV models will lead to better performance, is taking place. In this paper, we propose a novel multi-model and multi-level fusion strategy to tackle the SASV task. Compared with purely scoring fusion and embedding fusion methods, this framework first utilizes embeddings from CM models, propagating CM embeddings into a CM block to obtain a CM score. In the second-level fusion, the CM score and ASV scores directly from ASV systems will be concatenated into a prediction block for the final decision. As a result, the best single fusion system has achieved the SASV-EER of 0.97% on the evaluation set. Then by ensembling the top-5 fusion systems, the final SASV-EER reached 0.89%.

cs.SD↗

arXiv2022

One-loop Matching of the Type-II Seesaw Model onto the Standard Model Effective Field Theory

In this paper, we continue to construct the low-energy effective field theories (EFTs) of the canonical seesaw models, which are natural extensions of the Standard Model (SM) to accommodate tiny but nonzero neutrino masses. Different from three right-handed neutrino singlets in the type-I seesaw model, the Higgs triplet in the type-II seesaw model participates directly in the electroweak gauge interactions, rendering the EFT construction more challenging. By integrating out the heavy Higgs triplet in the functional-integral formalism, we carry out a complete one-loop matching of the type-II seesaw model onto the so-called Standard Model Effective Field Theory (SMEFT). It turns out that 41 dimension-six operators (barring flavor structures and Hermitian conjugates) in the Warsaw basis of the SMEFT can be obtained, covering all those 31 dimension-six operators in the case of type-I seesaw model. The Wilson coefficients for 41 dimension-six operators are computed up to $\mathcal{O}\left( M^{-2}_Δ\right)$ with $M^{}_Δ$ being the mass scale of the Higgs triplet. Moreover, the branching ratios of rare radiative decays of charged leptons $l^-_α\to l^-_β+ γ$ are calculated in the EFT and compared with that in the full theory in order to demonstrate the practical application and the correctness of our EFT construction.

hep-ph↗

arXiv2022

PathSAGE: Spatial Graph Attention Neural Networks With Random Path Sampling

Graph Convolutional Networks (GCNs) achieve great success in non-Euclidean structure data processing recently. In existing studies, deeper layers are used in CCNs to extract deeper features of Euclidean structure data. However, for non-Euclidean structure data, too deep GCNs will confront with problems like "neighbor explosion" and "over-smoothing", it also cannot be applied to large datasets. To address these problems, we propose a model called PathSAGE, which can learn high-order topological information and improve the model's performance by expanding the receptive field. The model randomly samples paths starting from the central node and aggregates them by Transformer encoder. PathSAGE has only one layer of structure to aggregate nodes which avoid those problems above. The results of evaluation shows that our model achieves comparable performance with the state-of-the-art models in inductive learning tasks.

cs.LG↗

arXiv2021

Decentralized Spectrum Access System: Vision, Challenges, and a Blockchain Solution

Spectrum access system (SAS) is widely considered the de facto solution to coordinating dynamic spectrum sharing (DSS) and protecting incumbent users. The current SAS paradigm prescribed by the FCC for the CBRS band and standardized by the WInnForum follows a centralized service model in that a spectrum user subscribes to a SAS server for spectrum allocation service. This model, however, neither tolerates SAS server failures (crash or Byzantine) nor resists dishonest SAS administrators, leading to serious concerns on SAS system reliability and trustworthiness. This is especially concerning for the evolving DSS landscape where an increasing number of SAS service providers and heterogeneous user requirements are coming up. To address these challenges, we propose a novel blockchain-based decentralized SAS architecture called BD-SAS that provides SAS services securely and efficiently, without relying on the trust of each individual SAS server for the overall system trustworthiness. In BD-SAS, a global blockchain (G-Chain) is used for spectrum regulatory compliance while smart contract-enabled local blockchains (L-Chains) are instantiated in individual spectrum zones for automating spectrum access assignment per user request. We hope our vision of a decentralized SAS, the BD-SAS architecture, and discussion on future challenges can open up a new direction towards reliable spectrum management in a decentralized manner.

cs.DC↗

arXiv2021

Longer time simulation of the unsteady Navier-Stokes equations based on a modified convective formulation

For the discretization of the convective term in the Navier-Stokes equations (NSEs), the commonly used convective formulation (CONV) does not preserve the energy if the divergence constraint is only weakly enforced. In this paper, we apply the skew-symmetrization technique in [B. Cockburn, G. Kanschat and D. Schötzau, Math. Comp., 74 (2005), pp. 1067-1095] to conforming finite element methods, which restores energy conservation for CONV. The crucial idea is to replace the discrete advective velocity with its a $H(\operatorname{div})$-conforming divergence-free approximation in CONV. We prove that the modified convective formulation also conserves linear momentum, helicity, 2D enstrophy and total vorticity under some appropriate senses. Its a Picard-type linearization form also conserves them. Under the assumption $\boldsymbol{u}\in L^{2}(0,T;\boldsymbol{W}^{1,\infty}(Ω)),$ it can be shown that the Gronwall constant does not explicitly depend on the Reynolds number in the error estimates. The long time numerical simulations show that the linearized and modified convective formulation has a similar performance with the EMAC formulation and outperforms the usual skew-symmetric formulation (SKEW).

math.NA↗

arXiv2021

Arbitrary Distribution Modeling with Censorship in Real-Time Bidding Advertising

The purpose of Inventory Pricing is to bid the right prices to online ad opportunities, which is crucial for a Demand-Side Platform (DSP) to win advertising auctions in Real-Time Bidding (RTB). In the planning stage, advertisers need the forecast of probabilistic models to make bidding decisions. However, most of the previous works made strong assumptions on the distribution form of the winning price, which reduced their accuracy and weakened their ability to make generalizations. Though some works recently tried to fit the distribution directly, their complex structure lacked efficiency on online inference. In this paper, we devise a novel loss function, Neighborhood Likelihood Loss (NLL), collaborating with a proposed framework, Arbitrary Distribution Modeling (ADM), to predict the winning price distribution under censorship with no pre-assumption required. We conducted experiments on two real-world experimental datasets and one large-scale, non-simulated production dataset in our system. Experiments showed that ADM outperformed the baselines both on algorithm and business metrics. By replaying historical data of the production environment, this method was shown to lead to good yield in our system. Without any pre-assumed specific distribution form, ADM showed significant advantages in effectiveness and efficiency, demonstrating its great capability in modeling sophisticated price landscapes.

cs.LG↗

arXiv2021

An EMA-conserving, pressure-robust and Re-semi-robust reconstruction method for the unsteady incompressible Navier-Stokes equations

Proper EMA-balance (E: kinetic energy; M: momentum; A: angular momentum), pressure-robustness and $Re$-semi-robustness ($Re$: Reynolds number) are three important properties of Navier-Stokes simulations with exactly divergence-free elements. This EMA-balance makes a method conserve kinetic energy, linear momentum and angular momentum under some suitable senses; pressure-robustness means that the velocity errors are independent of the continuous pressure; $Re$-semi-robustness means that the constants appearing in the error bounds of kinetic and dissipation energies do not explicitly depend on inverse powers of the viscosity. In this paper, based on the pressure-robust reconstruction methods in [{A. Linke and C. Merdon, {\it Comput. Methods Appl. Mech. Engrg.} 311 (2016), 304-326}], we propose a novel reconstruction method for a class of non-divergence-free simplicial elements which admits almost all the above properties. The only exception is the energy balance, where kinetic energy should be replaced by a properly redefined discrete energy. Some numerical comparisons with exactly divergence-free methods, pressure-robust reconstructions and the EMAC scheme are provided to confirm our theoretical results.

math.NA↗

arXiv2021

A low order divergence-free H(div)-conforming finite element method for Stokes flows

In this paper, we propose a ${ P_{1}^{c}}\oplus {RT0}-P0$ discretization of the Stokes equations on general simplicial meshes in two/three dimensions (2D/3D), which yields an exactly divergence-free and pressure-independent velocity approximation with optimal order. Our method has the following features. Firstly, the global number of the degrees of freedom of our method is the same as the low order Bernardi and Raugel ($B$-$R$) finite element method (Bernardi and Raugel, 1985), while the number of {the non-zero entries} of the former is about half of the latter in the velocity-velocity region of the coefficient matrix. Secondly, the ${ P_{1}^{c}}$ component of the velocity, the $RT0$ component of the velocity and the pressure seem to solve a popular ${ P_{1}^{c}}-{RT0}-P0$ discretization of a poroelastic-type system formally. Finally, our method can be easily transformed into a pressure-robust and stabilized ${ P_{1}^{c}}-P0$ discretization for the Stokes problem via the static condensation of the $RT0$ component, which has a much smaller number of global degrees of freedom. Numerical experiments illustrating the robustness of our method are also provided.

math.NA↗

arXiv2021

Positivity in Multi-Field EFTs

We discuss the general method for obtaining full positivity bounds on multi-field effective field theories (EFTs). While the leading order forward positivity bounds are commonly derived from the elastic scattering of two (superposed) external states, we show that for a generic EFT containing 3 or more low-energy modes, this approach only gives incomplete bounds. We then identify the allowed parameter space as the dual to a spectrahedron, constructed from crossing symmetries of the amplitude, and show that finding the optimal bounds for a given number of modes is equivalent to a geometric problem: finding the extremal rays of a spectrahedron. We show how this is done analytically for simple cases, and numerically formulated as semidefinite programming (SDP) problems for more complicated cases. We demonstrate this approach with a number of well-motivated examples in particle physics and cosmology, including EFTs of scalars, vectors, fermions and gravitons. In all these cases, we find that the SDP approach leads to results that either improve the previous ones or are completely new. We also find that the SDP approach is numerically much more efficient.

hep-ph↗

arXiv2021

S$^2$-MLPv2: Improved Spatial-Shift MLP Architecture for Vision

Recently, MLP-based vision backbones emerge. MLP-based vision architectures with less inductive bias achieve competitive performance in image recognition compared with CNNs and vision Transformers. Among them, spatial-shift MLP (S$^2$-MLP), adopting the straightforward spatial-shift operation, achieves better performance than the pioneering works including MLP-mixer and ResMLP. More recently, using smaller patches with a pyramid structure, Vision Permutator (ViP) and Global Filter Network (GFNet) achieve better performance than S$^2$-MLP. In this paper, we improve the S$^2$-MLP vision backbone. We expand the feature map along the channel dimension and split the expanded feature map into several parts. We conduct different spatial-shift operations on split parts. Meanwhile, we exploit the split-attention operation to fuse these split parts. Moreover, like the counterparts, we adopt smaller-scale patches and use a pyramid structure for boosting the image recognition accuracy. We term the improved spatial-shift MLP vision backbone as S$^2$-MLPv2. Using 55M parameters, our medium-scale model, S$^2$-MLPv2-Medium achieves an $83.6\%$ top-1 accuracy on the ImageNet-1K benchmark using $224\times 224$ images without self-attention and external training data.

cs.CV↗

arXiv2021

Channel-wise Gated Res2Net: Towards Robust Detection of Synthetic Speech Attacks

Existing approaches for anti-spoofing in automatic speaker verification (ASV) still lack generalizability to unseen attacks. The Res2Net approach designs a residual-like connection between feature groups within one block, which increases the possible receptive fields and improves the system's detection generalizability. However, such a residual-like connection is performed by a direct addition between feature groups without channel-wise priority. We argue that the information across channels may not contribute to spoofing cues equally, and the less relevant channels are expected to be suppressed before adding onto the next feature group, so that the system can generalize better to unseen attacks. This argument motivates the current work that presents a novel, channel-wise gated Res2Net (CG-Res2Net), which modifies Res2Net to enable a channel-wise gating mechanism in the connection between feature groups. This gating mechanism dynamically selects channel-wise features based on the input, to suppress the less relevant channels and enhance the detection generalizability. Three gating mechanisms with different structures are proposed and integrated into Res2Net. Experimental results conducted on ASVspoof 2019 logical access (LA) demonstrate that the proposed CG-Res2Net significantly outperforms Res2Net on both the overall LA evaluation set and individual difficult unseen attacks, which also outperforms other state-of-the-art single systems, depicting the effectiveness of our method.

eess.AS↗

arXiv2021

VAENAR-TTS: Variational Auto-Encoder based Non-AutoRegressive Text-to-Speech Synthesis

This paper describes a variational auto-encoder based non-autoregressive text-to-speech (VAENAR-TTS) model. The autoregressive TTS (AR-TTS) models based on the sequence-to-sequence architecture can generate high-quality speech, but their sequential decoding process can be time-consuming. Recently, non-autoregressive TTS (NAR-TTS) models have been shown to be more efficient with the parallel decoding process. However, these NAR-TTS models rely on phoneme-level durations to generate a hard alignment between the text and the spectrogram. Obtaining duration labels, either through forced alignment or knowledge distillation, is cumbersome. Furthermore, hard alignment based on phoneme expansion can degrade the naturalness of the synthesized speech. In contrast, the proposed model of VAENAR-TTS is an end-to-end approach that does not require phoneme-level durations. The VAENAR-TTS model does not contain recurrent structures and is completely non-autoregressive in both the training and inference phases. Based on the VAE architecture, the alignment information is encoded in the latent variable, and attention-based soft alignment between the text and the latent variable is used in the decoder to reconstruct the spectrogram. Experiments show that VAENAR-TTS achieves state-of-the-art synthesis quality, while the synthesis speed is comparable with other NAR-TTS models.

cs.SD↗