arXiv subjects

Zhenyu Zhou

Publications and source records attributed to Zhenyu Zhou.

At least 37 records · Page 2Linked to original sources

arXiv2024

A Comprehensive Investigation on Speaker Augmentation for Speaker Recognition

Data augmentation (DA) has played a pivotal role in the success of deep speaker recognition. Current DA techniques primarily focus on speaker-preserving augmentation, which does not change the speaker trait of the speech and does not create new speakers. Recent research has shed light on the potential of speaker augmentation, which generates new speakers to enrich the training dataset. In this study, we delve into two speaker augmentation approaches: speed perturbation (SP) and vocal tract length perturbation (VTLP). Despite the empirical utilization of both methods, a comprehensive investigation into their efficacy is lacking. Our study, conducted using two public datasets, VoxCeleb and CN-Celeb, revealed that both SP and VTLP are proficient at generating new speakers, leading to significant performance improvements in speaker recognition. Furthermore, they exhibit distinct properties in sensitivity to perturbation factors and data complexity, hinting at the potential benefits of their fusion. Our research underscores the substantial potential of speaker augmentation, highlighting the importance of in-depth exploration and analysis.

cs.SD↗

arXiv2024

On the Trajectory Regularity of ODE-based Diffusion Sampling

Diffusion-based generative models use stochastic differential equations (SDEs) and their equivalent ordinary differential equations (ODEs) to establish a smooth connection between a complex data distribution and a tractable prior distribution. In this paper, we identify several intriguing trajectory properties in the ODE-based sampling process of diffusion models. We characterize an implicit denoising trajectory and discuss its vital role in forming the coupled sampling trajectory with a strong shape regularity, regardless of the generated content. We also describe a dynamic programming-based scheme to make the time schedule in sampling better fit the underlying trajectory structure. This simple strategy requires minimal modification to any given ODE-based numerical solvers and incurs negligible computational cost, while delivering superior performance in image generation, especially in $5\sim 10$ function evaluations.

cs.LG↗

arXiv2024

Adversarial Data Augmentation for Robust Speaker Verification

Data augmentation (DA) has gained widespread popularity in deep speaker models due to its ease of implementation and significant effectiveness. It enriches training data by simulating real-life acoustic variations, enabling deep neural networks to learn speaker-related representations while disregarding irrelevant acoustic variations, thereby improving robustness and generalization. However, a potential issue with the vanilla DA is augmentation residual, i.e., unwanted distortion caused by different types of augmentation. To address this problem, this paper proposes a novel approach called adversarial data augmentation (A-DA) which combines DA with adversarial learning. Specifically, it involves an additional augmentation classifier to categorize various augmentation types used in data augmentation. This adversarial learning empowers the network to generate speaker embeddings that can deceive the augmentation classifier, making the learned speaker embeddings more robust in the face of augmentation variations. Experiments conducted on VoxCeleb and CN-Celeb datasets demonstrate that our proposed A-DA outperforms standard DA in both augmentation matched and mismatched test conditions, showcasing its superior robustness and generalization against acoustic variations.

cs.SD↗

arXiv2023

Full Conformal Oscillator Representations of Orthogonal Lie Algebras and Combinatorial Identities

Zhao and the second author (2013) constructed a functor from o(k)-Mod to o(k + 2)-Mod. In this paper, we use the functor successively to obtain an universal first-order differential operator realization for any highest-weight representation of o(2n + 3) in (n + 1)^2 variables and that of o(2n + 2) in n(n + 1) variables. When the highest weight is dominant integral, we determine the corresponding finite-dimensional irreducible module explicitly. One can use the result to study tensor decompositions of finite-dimensional irreducible modules by solving certain first-order linear partial differential equations, and thereby obtain the corresponding physically interested Clebsch-Gordan coefficients and exact solutions of Knizhnik-Zamolodchikov equation in WZW model of conformal field theory. We also find an equation of counting the dimension of an irreducible o(k + 2)-module in terms of certain alternating sum of the dimensions of irreducible o(k)-modules. In the case of the Steinberg modules, we obtain new combinatorial identities of classical type.

math.RT↗

arXiv2023

Scalable Electrodeposition of Eutectic Indium Gallium from an Acetonitrile-Based Electrolyte for Integrated Stretchable Electronics

For the advancement of highly-integrated stretchable electronics, the development of scalable sub-micrometer conductor patterning is required. Eutectic gallium indium EGaIn is an attractive conductor for stretchable electronics, as its liquid metallic character grants it high electrical conductivity upon deformation. However, its high surface energy precludes patterning it with (sub)-micron resolution. Herein, we overcome this limitation by reporting for the first time the electrodeposition of EGaIn. We use a non-aqueous acetonitrile-based electrolyte that exhibits high electrochemical stability and chemical orthogonality. The electrodeposited material led to low-resistance lines that remained stable upon (repeated) stretching to a 100 percent strain. Because electrodeposition benefits from the resolution of mature nanofabrication methods used to pattern the base metal, the proposed bottom-up approach achieved a record-high density integration of EGaIn regular lines of 300 nm half-pitch on an elastomer substrate by plating on a gold seed layer pre-patterned by nanoimprinting. Moreover, vertical integration was enabled by filling high aspect ratio vias. This capability was conceptualized by the fabrication of an omnidirectionally stretchable 3D electronic circuit, and demonstrates a soft-electronic analogue of the stablished damascene process used to fabricate microchip interconnects. Overall, this work proposes a simple route to address the challenge of metallization in highly integrated (3D) stretchable electronics.

physics.app-ph↗

arXiv2023

An Investigation of Distribution Alignment in Multi-Genre Speaker Recognition

Multi-genre speaker recognition is becoming increasingly popular due to its ability to better represent the complexities of real-world applications. However, a major challenge is the significant shift in the distribution of speaker vectors across different genres. While distribution alignment is a common approach to address this challenge, previous studies have mainly focused on aligning a source domain with a target domain, and the performance of multi-genre data is unknown. This paper presents a comprehensive study of mainstream distribution alignment methods on multi-genre data, where multiple distributions need to be aligned. We analyze various methods both qualitatively and quantitatively. Our experiments on the CN-Celeb dataset show that within-between distribution alignment (WBDA) performs relatively better. However, we also found that none of the investigated methods consistently improved performance in all test cases. This suggests that solely aligning the distributions of speaker vectors may not fully address the challenges posed by multi-genre speaker recognition. Further investigation is necessary to develop a more comprehensive solution.

cs.SD↗

arXiv2023

HDR Video Reconstruction with a Large Dynamic Dataset in Raw and sRGB Domains

High dynamic range (HDR) video reconstruction is attracting more and more attention due to the superior visual quality compared with those of low dynamic range (LDR) videos. The availability of LDR-HDR training pairs is essential for the HDR reconstruction quality. However, there are still no real LDR-HDR pairs for dynamic scenes due to the difficulty in capturing LDR-HDR frames simultaneously. In this work, we propose to utilize a staggered sensor to capture two alternate exposure images simultaneously, which are then fused into an HDR frame in both raw and sRGB domains. In this way, we build a large scale LDR-HDR video dataset with 85 scenes and each scene contains 60 frames. Based on this dataset, we further propose a Raw-HDRNet, which utilizes the raw LDR frames as inputs. We propose a pyramid flow-guided deformation convolution to align neighboring frames. Experimental results demonstrate that 1) the proposed dataset can improve the HDR reconstruction performance on real scenes for three benchmark networks; 2) Compared with sRGB inputs, utilizing raw inputs can further improve the reconstruction quality and our proposed Raw-HDRNet is a strong baseline for raw HDR reconstruction. Our dataset and code will be released after the acceptance of this paper.

eess.IV↗

arXiv2023

BEVFusion4D: Learning LiDAR-Camera Fusion Under Bird's-Eye-View via Cross-Modality Guidance and Temporal Aggregation

Integrating LiDAR and Camera information into Bird's-Eye-View (BEV) has become an essential topic for 3D object detection in autonomous driving. Existing methods mostly adopt an independent dual-branch framework to generate LiDAR and camera BEV, then perform an adaptive modality fusion. Since point clouds provide more accurate localization and geometry information, they could serve as a reliable spatial prior to acquiring relevant semantic information from the images. Therefore, we design a LiDAR-Guided View Transformer (LGVT) to effectively obtain the camera representation in BEV space and thus benefit the whole dual-branch fusion system. LGVT takes camera BEV as the primitive semantic query, repeatedly leveraging the spatial cue of LiDAR BEV for extracting image features across multiple camera views. Moreover, we extend our framework into the temporal domain with our proposed Temporal Deformable Alignment (TDA) module, which aims to aggregate BEV features from multiple historical frames. Including these two modules, our framework dubbed BEVFusion4D achieves state-of-the-art results in 3D object detection, with 72.0% mAP and 73.5% NDS on the nuScenes validation set, and 73.3% mAP and 74.7% NDS on nuScenes test set, respectively.

cs.CV↗

arXiv2022

Full Projective Oscillator Representations of Special Linear Lie Algebras and Combinatorial Identities

Using the projective oscillator representation of sl(n+1) and Shen's mixed product for Witt algebras, Zhao and the second author (2011) constructed a new functor from sl(n)-Mod to sl(n+1)-Mod. In this paper, we start from n = 2 and use the functor successively to obtain a full projective oscillator realization of any finite-dimensional irreducible representation of sl(n+1). The representation formulas of all the root vectors of sl(n+1) are given in terms of first-order differential operators in n(n+1)/2 variables. One can use the result to study tensor decompositions of finite-dimensional irreducible modules by solving certain first-order linear partial differential equations, and thereby obtain the corresponding physically interested Clebsch-Gordan coefficients and exact solutions of Knizhnik-Zamolodchikov equation in WZW model of conformal field theory.

math.RT↗

arXiv2022

Two-timescale Resource Allocation for Automated Networks in IIoT

The rapid technological advances of cellular technologies will revolutionize network automation in industrial internet of things (IIoT). In this paper, we investigate the two-timescale resource allocation problem in IIoT networks with hybrid energy supply, where temporal variations of energy harvesting (EH), electricity price, channel state, and data arrival exhibit different granularity. The formulated problem consists of energy management at a large timescale, as well as rate control, channel selection, and power allocation at a small timescale. To address this challenge, we develop an online solution to guarantee bounded performance deviation with only causal information. Specifically, Lyapunov optimization is leveraged to transform the long-term stochastic optimization problem into a series of short-term deterministic optimization problems. Then, a low-complexity rate control algorithm is developed based on alternating direction method of multipliers (ADMM), which accelerates the convergence speed via the decomposition-coordination approach. Next, the joint channel selection and power allocation problem is transformed into a one-to-many matching problem, and solved by the proposed price-based matching with quota restriction. Finally, the proposed algorithm is verified through simulations under various system configurations.

eess.SP↗

arXiv2021

Molecular modeling with machine-learned universal potential functions

Molecular modeling is an important topic in drug discovery. Decades of research have led to the development of high quality scalable molecular force fields. In this paper, we show that neural networks can be used to train a universal approximator for energy potential functions. By incorporating a fully automated training process we have been able to train smooth, differentiable, and predictive potential functions on large-scale crystal structures. A variety of tests have also been performed to show the superiority and versatility of the machine-learned model.

q-bio.QM↗

arXiv2020

Boosting Gradient for White-Box Adversarial Attacks

Deep neural networks (DNNs) are playing key roles in various artificial intelligence applications such as image classification and object recognition. However, a growing number of studies have shown that there exist adversarial examples in DNNs, which are almost imperceptibly different from original samples, but can greatly change the network output. Existing white-box attack algorithms can generate powerful adversarial examples. Nevertheless, most of the algorithms concentrate on how to iteratively make the best use of gradients to improve adversarial performance. In contrast, in this paper, we focus on the properties of the widely-used ReLU activation function, and discover that there exist two phenomena (i.e., wrong blocking and over transmission) misleading the calculation of gradients in ReLU during the backpropagation. Both issues enlarge the difference between the predicted changes of the loss function from gradient and corresponding actual changes, and mislead the gradients which results in larger perturbations. Therefore, we propose a universal adversarial example generation method, called ADV-ReLU, to enhance the performance of gradient based white-box attack algorithms. During the backpropagation of the network, our approach calculates the gradient of the loss function versus network input, maps the values to scores, and selects a part of them to update the misleading gradients. Comprehensive experimental results on \emph{ImageNet} demonstrate that our ADV-ReLU can be easily integrated into many state-of-the-art gradient-based white-box attack algorithms, as well as transferred to black-box attack attackers, to further decrease perturbations in the ${\ell _2}$-norm.

cs.LG↗

arXiv2019

Probe Charm Yukawa at the future $e^{-}p$ and $e^{+}e^{-}$ colliders

Large Hadron Collider (LHC) has provided direct evidence of Yukawa couplings between the third generation charged fermions and the 125 GeV Higgs boson. Whether the first two generation charged fermions arise from exactly the same mechanism becomes the next interesting question. Therefore, direct measurements of charm or muon Yukawa couplings will be crucial to answering this puzzle. The charm Yukawa measurement at the LHC suffers from severe QCD background and it is extremely difficult to reach the sensitivity. In this paper, we compare the potential of probing charm Yukawa coupling at the two proposed future "Higgs Factory" experiments, the Large Hadron electron Collider (LHeC) and Circular electron positron collider (CEPC). At the LHeC, Higgs bosons will be produced via weak boson fusion and the energetic forward jet may suppress the background significantly. However, due to huge $γ-g$ scattering background, the potential of LHeC search is still limited. With $-80\%$ polarized electron beam of 60 GeV, the signal significance can only reach $2σ$ for $ κ_{c}\simeq 1$ with a 3 ab$^{-1}$ integrated luminosity. In comparison, measurement at the CEPC can then reach $8.0σ$ for $κ_{c}\simeq 1$ with a 2 ab$^{-1}$ of data.

hep-ph↗

arXiv2019

When Internet of Things Meets Blockchain: Challenges in Distributed Consensus

Blockchain has been regarded as a promising technology for Internet of Things (IoT), since it provides significant solutions for decentralized network which can address trust and security concerns, high maintenance cost problem, etc. The decentralization provided by blockchain can be largely attributed to the use of consensus mechanism, which enables peer-to-peer trading in a distributed manner without the involvement of any third party. This article starts from introducing the basic concept of blockchain and illustrating why consensus mechanism plays an indispensable role in a blockchain enabled IoT system. Then, we discuss the main ideas of two famous consensus mechanisms including Proof of Work (PoW) and Proof of Stake (PoS), and list their limitations in IoT. Next, two mainstream Direct Acyclic Graph (DAG) based consensus mechanisms, i.e., the Tangle and Hashgraph, are reviewed to show why DAG consensus is more suitable for IoT system than PoW and PoS. Potential issues and challenges of DAG based consensus mechanism to be addressed in the future are discussed in the last.

cs.DC↗

arXiv2018

A Low-latency Secure Data Outsourcing Scheme for Cloud-WSN

With the support of cloud computing, large quantities of data collected from various WSN applications can be managed efficiently. However, maintaining data security and efficiency of data processing in cloud-WSN (C-WSN) are important and challenging issues. In this paper, we present an efficient data outsourcing scheme based on CP-ABE, which can not only guarantee secure data access, but also reduce overall data processing time. In our proposed scheme, a large file is divided into several data blocks by data owner (DO) firstly. Then, the data blocks are encrypted and transferred to the cloud server in parallel. For data receiver (DR), data decryption and data transmission is also processed in parallel. In addition, data integrity can be checked by DR without any master key components. The security analysis shows that the proposed scheme can meet the security requirement of C-WSN. By performance evaluation, it shows that our scheme can dramatically improve data processing efficiency compared to the traditional CP-ABE method.

cs.CR↗

arXiv2017

Prediction of amino acid side chain conformation using a deep neural network

A deep neural network based architecture was constructed to predict amino acid side chain conformation with unprecedented accuracy. Amino acid side chain conformation prediction is essential for protein homology modeling and protein design. Current widely-adopted methods use physics-based energy functions to evaluate side chain conformation. Here, using a deep neural network architecture without physics-based assumptions, we have demonstrated that side chain conformation prediction accuracy can be improved by more than 25%, especially for aromatic residues compared with current standard methods. More strikingly, the prediction method presented here is robust enough to identify individual conformational outliers from high resolution structures in a protein data bank without providing its structural factors. We envisage that our amino acid side chain predictor could be used as a quality check step for future protein structure model validation and many other potential applications such as side chain assignment in Cryo-electron microscopy, crystallography model auto-building, protein folding and small molecule ligand docking.

q-bio.BM↗

arXiv2016

Policy Optimization for Content Push via Energy Harvesting Small Cells in Heterogeneous Networks

Motivated by the rapid development of energy harvesting technology and content-aware communication in access networks, this paper considers the push mechanism design in small-cell base stations (SBSs) powered by renewable energy. A user request can be satisfied by either push or unicast from the SBS. If the SBS cannot handle the request, the user is blocked by the SBS and is served by the macro-cell BS (MBS) instead, which typically consumes more energy. We aim to minimize the ratio of user requests blocked by the SBS. With finite battery capacity, Markov decision process based problem is formulated, and the optimal policy is found by dynamic programming (DP). Two threshold-based policies are proposed: the push-only threshold-based (POTB) policy and the energy-efficient threshold-based (EETB) policy, and the closed-form blocking probabilities with infinite battery capacity are derived. Numerical results show that the proposed policies outperform the conventional non-push policy if the content popularity changes slowly or the content request generating rate is high, and can achieve the performance of the greedy optimal threshold-based (GOTB) policy. In addition, the performance gap between the threshold-based policies and the DP optimal policy is small when the energy arrival rate is low or the request generating rate is high.

cs.IT↗

arXiv2016

Networked MIMO with Fractional Joint Transmission in Energy Harvesting Systems

This paper considers two base stations (BSs) powered by renewable energy serving two users cooperatively. With different BS energy arrival rates, a fractional joint transmission (JT) strategy is proposed, which divides each transmission frame into two subframes. In the first subframe, one BS keeps silent to store energy while the other transmits data, and then they perform zero-forcing JT (ZF-JT) in the second subframe. We consider the average sum-rate maximization problem by optimizing the energy allocation and the time fraction of ZF-JT in two steps. Firstly, the sum-rate maximization for given energy budget in each frame is analyzed. We prove that the optimal transmit power can be derived in closed-form, and the optimal time fraction can be found via bi-section search. Secondly, approximate dynamic programming (DP) algorithm is introduced to determine the energy allocation among frames. We adopt a linear approximation with the features associated with system states, and determine the weights of features by simulation. We also operate the approximation several times with random initial policy, named as policy exploration, to broaden the policy search range. Numerical results show that the proposed fractional JT greatly improves the performance. Also, appropriate policy exploration is shown to perform close to the optimal.

cs.IT↗