arXiv subjects

Yu Zhang

Publications and source records attributed to Yu Zhang.

At least 1,333 records · Page 74Linked to original sources

arXiv2020

WaveGrad: Estimating Gradients for Waveform Generation

This paper introduces WaveGrad, a conditional model for waveform generation which estimates gradients of the data density. The model is built on prior work on score matching and diffusion probabilistic models. It starts from a Gaussian white noise signal and iteratively refines the signal via a gradient-based sampler conditioned on the mel-spectrogram. WaveGrad offers a natural way to trade inference speed for sample quality by adjusting the number of refinement steps, and bridges the gap between non-autoregressive and autoregressive models in terms of audio quality. We find that it can generate high fidelity audio samples using as few as six iterations. Experiments reveal WaveGrad to generate high fidelity audio, outperforming adversarial non-autoregressive baselines and matching a strong likelihood-based autoregressive baseline using fewer sequential operations. Audio samples are available at https://wavegrad.github.io/.

eess.AS↗

arXiv2020

A Survey of Non-Volatile Main Memory Technologies: State-of-the-Arts, Practices, and Future Directions

Non-Volatile Main Memories (NVMMs) have recently emerged as promising technologies for future memory systems. Generally, NVMMs have many desirable properties such as high density, byte-addressability, non-volatility, low cost, and energy efficiency, at the expense of high write latency, high write power consumption and limited write endurance. NVMMs have become a competitive alternative of Dynamic Random Access Memory (DRAM), and will fundamentally change the landscape of memory systems. They bring many research opportunities as well as challenges on system architectural designs, memory management in operating systems (OSes), and programming models for hybrid memory systems. In this article, we first revisit the landscape of emerging NVMM technologies, and then survey the state-of-the-art studies of NVMM technologies. We classify those studies with a taxonomy according to different dimensions such as memory architectures, data persistence, performance improvement, energy saving, and wear leveling. Second, to demonstrate the best practices in building NVMM systems, we introduce our recent work of hybrid memory system designs from the dimensions of architectures, systems, and applications. At last, we present our vision of future research directions of NVMMs and shed some light on design challenges and opportunities.

cs.DC↗

arXiv2020

Cross-Modal Alignment with Mixture Experts Neural Network for Intral-City Retail Recommendation

In this paper, we introduce Cross-modal Alignment with mixture experts Neural Network (CameNN) recommendation model for intral-city retail industry, which aims to provide fresh foods and groceries retailing within 5 hours delivery service arising for the outbreak of Coronavirus disease (COVID-19) pandemic around the world. We propose CameNN, which is a multi-task model with three tasks including Image to Text Alignment (ITA) task, Text to Image Alignment (TIA) task and CVR prediction task. We use pre-trained BERT to generate the text embedding and pre-trained InceptionV4 to generate image patch embedding (each image is split into small patches with the same pixels and treat each patch as an image token). Softmax gating networks follow to learn the weight of each transformer expert output and choose only a subset of experts conditioned on the input. Then transformer encoder is applied as the share-bottom layer to learn all input features' shared interaction. Next, mixture of transformer experts (MoE) layer is implemented to model different aspects of tasks. At top of the MoE layer, we deploy a transformer layer for each task as task tower to learn task-specific information. On the real word intra-city dataset, experiments demonstrate CameNN outperform baselines and achieve significant improvements on the image and text representation. In practice, we applied CameNN on CVR prediction in our intra-city recommender system which is one of the leading intra-city platforms operated in China.

cs.CV↗

arXiv2020

Investigating Variable Stars in the Open Cluster NGC 1912 and Its Surrounding Field

In this work, we studied the variable stars in the open cluster NGC 1912 based on the photometric observations and $Gaia$ DR2 data. More than 3600 CCD frames in B, V, R filters were reduced, and we obtained the light curves that span about 63 hours. By analyzing these light curves, we detected 24 variable stars, including 16 periodic variable stars, seven eclipsing binaries and one star whose type is unclear. Among these 24 variable stars, 11 are newly-discovered, which are classified as six gamma Doradus stars, one delta Scuti star, three detached binaries and one contact binaries. We also confirmed 13 previously known variable stars. Based on cluster members identified by Cantat-Gaudin et al. (2018), we inferred cluster memberships for these detected variable stars. Using Gaia DR2 data, we plotted a new color-magnitude diagram for NGC 1912, and showed the nature of variable cluster members in kinematical properties and heliocentric distance. Among the 24 variable stars, seven variables are probable cluster members, which show homogeneity in kinematic characters and space position with the established cluster members. Four of the seven variable cluster members are the previously discovered stars, consisting of two gamma Dor stars and two delta Sct stars. The remaining three variable cluster members, which are all gamma Dor stars, are firstly detected in this work. The main physical parameters of these variable cluster members estimated from the color-magnitude diagram are log(age/yr)=8.75, [Fe/H]=-0.1, m-M=10.03 mag, and E(B-V)=0.307.

astro-ph.SR↗

arXiv2020

Correlation-Informed Permutation of Qubits for Reducing Ansatz Depth in VQE

The Variational Quantum Eigensolver (VQE) is a method of choice to solve the electronic structure problem for molecules on near-term gate-based quantum computers. However, the circuit depth is expected to grow significantly with problem size. Increased depth can both degrade the accuracy of the results and reduce trainability. In this work, we propose a novel approach to reduce ansatz circuit depth. Our approach, called PermVQE, adds an additional optimization loop to VQE that permutes qubits in order to solve for the qubit Hamiltonian that minimizes long-range correlations in the ground state. The choice of permutations is based on mutual information, which is a measure of interaction between electrons in spin-orbitals. Encoding strongly interacting spin-orbitals into proximal qubits on a quantum chip naturally reduces the circuit depth needed to prepare the ground state. For representative molecular systems, LiH, H$_2$, (H$_2$)$_2$, H$_4$, and H$_3^+$, we demonstrate for linear qubit connectivity that placing entangled qubits in close proximity leads to shallower depth circuits required to reach a given eigenvalue-eigenvector accuracy. This approach can be extended to any qubit connectivity and can significantly reduce the depth required to reach a desired accuracy in VQE. Moreover, our approach can be applied to other variational quantum algorithms beyond VQE.

quant-ph↗

arXiv2020

Improved Trainable Calibration Method for Neural Networks on Medical Imaging Classification

Recent works have shown that deep neural networks can achieve super-human performance in a wide range of image classification tasks in the medical imaging domain. However, these works have primarily focused on classification accuracy, ignoring the important role of uncertainty quantification. Empirically, neural networks are often miscalibrated and overconfident in their predictions. This miscalibration could be problematic in any automatic decision-making system, but we focus on the medical field in which neural network miscalibration has the potential to lead to significant treatment errors. We propose a novel calibration approach that maintains the overall classification accuracy while significantly improving model calibration. The proposed approach is based on expected calibration error, which is a common metric for quantifying miscalibration. Our approach can be easily integrated into any classification task as an auxiliary loss term, thus not requiring an explicit training round for calibration. We show that our approach reduces calibration error significantly across various architectures and datasets.

cs.CV↗

arXiv2020

Boosting Retailer Revenue by Generated Optimized Combined Multiple Digital Marketing Campaigns

Campaign is a frequently employed instrument in lifting up the GMV (Gross Merchandise Volume) of retailer in traditional marketing. As its counterpart in online context, digital-marketing-campaign (DMC) has being trending in recent years with the rapid development of the e-commerce. However, how to empower massive sellers on the online retailing platform the capacity of applying combined multiple digital marketing campaigns to boost their shops' revenue, is still a novel topic. In this work, a comprehensive solution of generating optimized combined multiple DMCs is presented. Firstly, a potential personalized DMC pool is generated for every retailer by a newly proposed neural network model, i.e. the DMCNet (Digital-Marketing-Campaign Net). Secondly, based on the sub-modular optimization theory and the DMC pool by DMCNet, the generated combined multiple DMCs are ranked with respect to their revenue generation strength then the top three ranked campaigns are returned to the sellers' back-end management system, so that retailers can set combined multiple DMCs for their online shops just in one-shot. Real online A/B-test shows that with the integrated solution, sellers of the online retailing platform increase their shops' GMVs with approximately 6$\%$.

cs.IR↗

arXiv2020

Effects of centrality fluctuation and deuteron formation on proton number cumulant in Au+Au collisions at $\sqrt{s_\mathrm{NN}}$ = 3 GeV from JAM model

We studied the effects of centrality fluctuation and deuteron formation on the cumulants ($C_n$) and correlation functions ($κ_n$) of protons up to sixth order in most central ($b<3$ fm) Au+Au collisions at $\sqrt{s_\mathrm{NN}}$ = 3 GeV from a microscopic transport model (JAM). The results are presented as a function of rapidity acceptance within transverse momentum $0.4<p_{T}<2 $ GeV/$c$. We compared the results obtained by centrality bin width correction (CBWC) using charged reference particle multiplicity with CBWC done using impact parameter bins. It was found that at low energies the centrality resolution for determining the collision centrality using charged particle multiplicities is not good enough to reduce the initial volume fluctuations effect for higher-order cumulant analysis. New methods need to be developed to classify events with high centrality resolution for heavy-ion collisions at low energies. Finally, we observed that the formation of deuteron will suppress the higher-order cumulants and correlation functions of protons and is found to be similar to the efficiency effect. This work can serve as a noncritical baseline for the QCD critical point search at the high baryon density region.

nucl-ex↗

arXiv2020

Online Explanation Generation for Human-Robot Teaming

As AI becomes an integral part of our lives, the development of explainable AI, embodied in the decision-making process of an AI or robotic agent, becomes imperative. For a robotic teammate, the ability to generate explanations to justify its behavior is one of the key requirements of explainable agency. Prior work on explanation generation has been focused on supporting the rationale behind the robot's decision or behavior. These approaches, however, fail to consider the mental demand for understanding the received explanation. In other words, the human teammate is expected to understand an explanation no matter how much information is presented. In this work, we argue that explanations, especially those of a complex nature, should be made in an online fashion during the execution, which helps spread out the information to be explained and thus reduce the mental workload of humans in highly cognitive demanding tasks. However, a challenge here is that the different parts of an explanation may be dependent on each other, which must be taken into account when generating online explanations. To this end, a general formulation of online explanation generation is presented with three variations satisfying different "online" properties. The new explanation generation methods are based on a model reconciliation setting introduced in our prior work. We evaluated our methods both with human subjects in a simulated rover domain, using NASA Task Load Index (TLX), and synthetically with ten different problems across two standard IPC domains. Results strongly suggest that our methods generate explanations that are perceived as less cognitively demanding and much preferred over the baselines and are computationally efficient.

cs.AI↗

arXiv2020

Better Than Reference In Low Light Image Enhancement: Conditional Re-Enhancement Networks

Low light images suffer from severe noise, low brightness, low contrast, etc. In previous researches, many image enhancement methods have been proposed, but few methods can deal with these problems simultaneously. In this paper, to solve these problems simultaneously, we propose a low light image enhancement method that can combined with supervised learning and previous HSV (Hue, Saturation, Value) or Retinex model based image enhancement methods. First, we analyse the relationship between the HSV color space and the Retinex theory, and show that the V channel (V channel in HSV color space, equals the maximum channel in RGB color space) of the enhanced image can well represent the contrast and brightness enhancement process. Then, a data-driven conditional re-enhancement network (denoted as CRENet) is proposed. The network takes low light images as input and the enhanced V channel as condition, then it can re-enhance the contrast and brightness of the low light image and at the same time reduce noise and color distortion. It should be noted that during the training process, any paired images with different exposure time can be used for training, and there is no need to carefully select the supervised images which will save a lot. In addition, it takes less than 20 ms to process a color image with the resolution 400*600 on a 2080Ti GPU. Finally, some comparative experiments are implemented to prove the effectiveness of the method. The results show that the method proposed in this paper can significantly improve the quality of the enhanced image, and by combining with other image contrast enhancement methods, the final enhancement result can even be better than the reference image in contrast and brightness. (Code will be available at https://github.com/hitzhangyu/image-enhancement-with-denoise)

eess.IV↗

arXiv2020

An End-to-End Attack on Text-based CAPTCHAs Based on Cycle-Consistent Generative Adversarial Network

As a widely deployed security scheme, text-based CAPTCHAs have become more and more difficult to resist machine learning-based attacks. So far, many researchers have conducted attacking research on text-based CAPTCHAs deployed by different companies (such as Microsoft, Amazon, and Apple) and achieved certain results.However, most of these attacks have some shortcomings, such as poor portability of attack methods, requiring a series of data preprocessing steps, and relying on large amounts of labeled CAPTCHAs. In this paper, we propose an efficient and simple end-to-end attack method based on cycle-consistent generative adversarial networks. Compared with previous studies, our method greatly reduces the cost of data labeling. In addition, this method has high portability. It can attack common text-based CAPTCHA schemes only by modifying a few configuration parameters, which makes the attack easier. Firstly, we train CAPTCHA synthesizers based on the cycle-GAN to generate some fake samples. Basic recognizers based on the convolutional recurrent neural network are trained with the fake data. Subsequently, an active transfer learning method is employed to optimize the basic recognizer utilizing tiny amounts of labeled real-world CAPTCHA samples. Our approach efficiently cracked the CAPTCHA schemes deployed by 10 popular websites, indicating that our attack is likely very general. Additionally, we analyzed the current most popular anti-recognition mechanisms. The results show that the combination of more anti-recognition mechanisms can improve the security of CAPTCHA, but the improvement is limited. Conversely, generating more complex CAPTCHAs may cost more resources and reduce the availability of CAPTCHAs.

cs.CV↗

arXiv2020

Twistronics in graphene-based van der Waals structures

The electronic properties of van der Waals (vdW) structures can be substantially modified by the moire superlattice potential, which strongly depends on the twist angle among the compounds. In twisted bilayer graphene (TBG), two low-energy Van Hove singularities (VHSs) move closer with decreasing twist angles and finally become highly non-dispersive flat bands at the magic angle (~ 1.1 degree). When the Fermi level lies within the flat bands of the TBG near the magic angle, Coulomb interaction is supposed to exceed the kinetic energy of the electrons, which can drive the system into various strongly correlated phases. Moreover, the strongly correlated states of flat bands are also realized in other graphene-based vdW structures with an interlayer twist. In this article, we mainly review the recent experimental advances on the strongly correlated physics of the magic-angle TBG (MATBG) and the small-angle twisted multilayer graphene. Lastly we will give out a perspective of this field.

cond-mat.mtrl-sci↗

arXiv2020

Location Information Aided Multiple Intelligent Reflecting Surface Systems

This paper proposes a novel location information aided multiple intelligent reflecting surface (IRS) systems. Assuming imperfect user location information, the effective angles from the IRS to the users are estimated, which is then used to design the transmit beam and IRS beam. Furthermore, closed-form expressions for the achievable rate are derived. The analytical findings indicate that the achievable rate can be improved by increasing the number of base station (BS) antennas or reflecting elements. Specifically, a power gain of order $N M^2$ is achieved, where $N$ is the antenna number and $M$ is the number of reflecting elements. Moreover, with a large number of reflecting elements, the individual signal to interference plus noise ratio (SINR) is proportional to $M$, while becomes proportional to $M^2$ as non-line-of-sight (NLOS) paths vanish. Also, it has been shown that high location uncertainty would significantly degrade the achievable rate. Besides, IRSs should be deployed at distinct directions (relative to the BS) and be far away from each other to reduce the interference from multiple IRSs. Finally, an optimal power allocation scheme has been proposed to improve the system performance.

cs.IT↗

arXiv2020

Channel Estimation and Hybrid Precoding for Distributed Phased Arrays Based MIMO Wireless Communications

Distributed phased arrays based multiple-input multiple-output (DPA-MIMO) is a newly introduced architecture that enables both spatial multiplexing and beamforming while facilitating highly reconfigurable hardware implementation in millimeter-wave (mmWave) frequency bands. With a DPA-MIMO system, we focus on channel state information (CSI) acquisition and hybrid precoding. As benefited from a coordinated and open-loop pilot beam pattern design, all the sub-arrays can perform channel sounding with less training overhead compared with the traditional orthogonal operation of each sub-array. Furthermore, two sparse channel recovery algorithms, known as joint orthogonal matching pursuit (JOMP) and joint sparse Bayesian learning with $\ell_2$ reweighting (JSBL-$\ell_2$), are proposed to exploit the hidden structured sparsity in the beam-domain channel vector. Finally, successive interference cancellation (SIC) based hybrid precoding through sub-array grouping is illustrated for the DPA-MIMO system, which decomposes the joint sub-array RF beamformer design into an interactive per-sub-array-group handle. Simulation results show that the proposed two channel estimators fully take advantage of the partial coupling characteristic of DPA-MIMO channels to perform channel recovery, and the proposed hybrid precoding algorithm is suitable for such array-of-sub-arrays architecture with satisfactory performance and low complexity.

cs.IT↗

arXiv2020

Is POS Tagging Necessary or Even Helpful for Neural Dependency Parsing?

In the pre deep learning era, part-of-speech tags have been considered as indispensable ingredients for feature engineering in dependency parsing. But quite a few works focus on joint tagging and parsing models to avoid error propagation. In contrast, recent studies suggest that POS tagging becomes much less important or even useless for neural parsing, especially when using character-based word representations. Yet there are not enough investigations focusing on this issue, both empirically and linguistically. To answer this, we design and compare three typical multi-task learning framework, i.e., Share-Loose, Share-Tight, and Stack, for joint tagging and parsing based on the state-of-the-art biaffine parser. Considering that it is much cheaper to annotate POS tags than parse trees, we also investigate the utilization of large-scale heterogeneous POS tag data. We conduct experiments on both English and Chinese datasets, and the results clearly show that POS tagging (both homogeneous and heterogeneous) can still significantly improve parsing performance when using the Stack joint framework. We conduct detailed analysis and gain more insights from the linguistic aspect.

cs.CL↗

arXiv2020

Fast and Accurate Neural CRF Constituency Parsing

Estimating probability distribution is one of the core issues in the NLP field. However, in both deep learning (DL) and pre-DL eras, unlike the vast applications of linear-chain CRF in sequence labeling tasks, very few works have applied tree-structure CRF to constituency parsing, mainly due to the complexity and inefficiency of the inside-outside algorithm. This work presents a fast and accurate neural CRF constituency parser. The key idea is to batchify the inside algorithm for loss computation by direct large tensor operations on GPU, and meanwhile avoid the outside algorithm for gradient computation via efficient back-propagation. We also propose a simple two-stage bracketing-then-labeling parsing approach to improve efficiency further. To improve the parsing performance, inspired by recent progress in dependency parsing, we introduce a new scoring architecture based on boundary representation and biaffine attention, and a beneficial dropout strategy. Experiments on PTB, CTB5.1, and CTB7 show that our two-stage CRF parser achieves new state-of-the-art performance on both settings of w/o and w/ BERT, and can parse over 1,000 sentences per second. We release our code at https://github.com/yzhangcs/crfpar.

cs.CL↗

arXiv2020

MiNet: Mixed Interest Network for Cross-Domain Click-Through Rate Prediction

Click-through rate (CTR) prediction is a critical task in online advertising systems. Existing works mainly address the single-domain CTR prediction problem and model aspects such as feature interaction, user behavior history and contextual information. Nevertheless, ads are usually displayed with natural content, which offers an opportunity for cross-domain CTR prediction. In this paper, we address this problem and leverage auxiliary data from a source domain to improve the CTR prediction performance of a target domain. Our study is based on UC Toutiao (a news feed service integrated with the UC Browser App, serving hundreds of millions of users daily), where the source domain is the news and the target domain is the ad. In order to effectively leverage news data for predicting CTRs of ads, we propose the Mixed Interest Network (MiNet) which jointly models three types of user interest: 1) long-term interest across domains, 2) short-term interest from the source domain and 3) short-term interest in the target domain. MiNet contains two levels of attentions, where the item-level attention can adaptively distill useful information from clicked news / ads and the interest-level attention can adaptively fuse different interest representations. Offline experiments show that MiNet outperforms several state-of-the-art methods for CTR prediction. We have deployed MiNet in UC Toutiao and the A/B test results show that the online CTR is also improved substantially. MiNet now serves the main ad traffic in UC Toutiao.

cs.IR↗

arXiv2020

Ultrafast triggering of insulator-metal transition in two-dimensional VSe$_2$

Assembling transition metal dichalcogenides (TMDCs) at the two-dimensional (2D) limit is a promising approach for tailoring emerging states of matter such as superconductivity or charge density waves (CDWs). Single-layer (SL) VSe$_2$ stands out in this regard because it exhibits a strongly enhanced CDW transition with a higher transition temperature compared to the bulk in addition to an insulating phase with an anisotropic gap at the Fermi level, causing a suppression of anticipated 2D ferromagnetism in the material. Here, we investigate the interplay of electronic and lattice degrees of freedom that underpin these electronic phases in SL VSe$_2$ using ultrafast pump-probe photoemission spectroscopy. In the insulating state, we observe a light-induced closure of the energy gap on a timescale of 480 fs, which we disentangle from the ensuing hot carrier dynamics. Our work thereby reveals that the phase transition in SL VSe$_2$ is driven by electron-lattice coupling and demonstrates the potential for controlling electronic phases in 2D materials with light.

cond-mat.mtrl-sci↗